# PLASTID PROTEOSTASIS: RELEVANCE OF TRANSCRIPTION, TRANSLATION AND POST-TRANSLATIONAL MODIFICATIONS

EDITED BY: Fiammetta Alagna, Michele Bellucci, Dario Leister and Andrea Pompa PUBLISHED IN: Frontiers in Plant Science

#### *Frontiers Copyright Statement*

*© Copyright 2007-2017 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.*

*The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.*

*Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.*

*Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.*

*As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.*

> *All copyright, and all rights therein, are protected by national and international copyright laws.*

*The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use.*

ISSN 1664-8714 ISBN 978-2-88945-343-6 DOI 10.3389/978-2-88945-343-6

### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# **PLASTID PROTEOSTASIS: RELEVANCE OF TRANSCRIPTION, TRANSLATION AND POST-TRANSLATIONAL MODIFICATIONS**

### Topic Editors:

**Fiammetta Alagna,** Consiglio per la Ricerca in Agricoltura e l'Analisi dell'Economia Agraria (CREA), Italy

**Michele Bellucci,** Consiglio Nazionale delle Ricerche (CNR), Italy **Dario Leister**, Ludwig-Maximilians-Universität München, Germany **Andrea Pompa,** Consiglio Nazionale delle Ricerche (CNR), Italy

An image of the adaxial side of a sugar beet leaf taken with a conventional fluorescence microscope. The green color is due to autofluorescence of chlorophyll in the chloroplasts. In the box, the optical section of a tobacco protoplast taken with a confocal microscope (at left) and a representation of the chloroplasts structure based on a transmission electron microscopy (TEM) image (at right) is shown. The drawings inside the TEM image represent translation of chloroplast mRNA by 70S ribosomes.

Image: Michele Bellucci.

Due to their bacterial endosymbiotic origin plastids are organelles with both nuclear-encoded and plastid-encoded proteins. Therefore, a highly integrated modulation of gene expression between the nucleus and the plastome is needed in plant cell development. Plastids have retained for the most part a prokaryotic gene expression machinery but, differently from prokaryotes and eukaryotes, they have largely abandoned transcriptional control and switched to predominantly translational control of their gene expression. Some transcriptional regulation is known to occur, but the coordinate expression between the nucleus and the plastome takes place mainly through translational regulation. However, the regulatory mechanisms of plastid gene expression (PGE) are mediated by intricate plastid-nuclear interactions and are still far from being fully understood. Although, for example, translational autoregulation mechanisms in algae have been described for subunits of heteromeric protein complexes and termed control by epistasy of synthesis (CES), only few autoregulatory proteins have been identified in plant plastids. It should be noted of course that PGE in C. reinhardtii is different from that in plants in many aspects. Another example of investigation in this research area is to understand the interactions that occur during RNA binding between nucleus-encoded RNA-binding proteins and the respective RNA sequences, and how this influences the translation initiation process.

In addition to this, the plastid retains a whole series of mechanisms for the preservation of its protein balance (proteostasis), including specific proteases, as well as molecular chaperones and enzymes useful in protein folding. After synthesis, plastid proteins must rapidly fold into stable three dimensional structures and often undergo co- and posttranslational modifications to perform their biological mission, avoiding aberrant folding, aggregation and targeting with the help of molecular chaperones and proteases.

We believe that this topic is highly interesting for many research areas because the regulation of PGE is not only of wide interest for plant biologists but has also biotechnological implications. Indeed, plastid transformation turns out to be a very promising tool for the production of recombinant proteins in plants, yet some limitations must still be overcome and we believe that this is mainly due to our limited knowledge of the mechanisms in plastids influencing the maintenance of proteostasis.

**Citation:** Alagna, F., Bellucci, M., Leister, D. , Pompa, A., eds. (2017). Plastid Proteostasis: Relevance of Transcription, Translation and Post-Translational Modifications. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-343-6

# Table of Contents

*05 Editorial: Plastid Proteostasis: Relevance of Transcription, Translation, and Post-translational Modifications*

Fiammetta Alagna, Michele Bellucci, Dario Leister and Andrea Pompa

### **Chapter 1. Transcriptional control of gene-expression**


Duorong Xu, Dario Leister and Tatjana Kleine

### **Chapter 2. Post-translational mechanisms affecting plastid proteostasis**

*46 Import of Soluble Proteins into Chloroplasts and Potential Regulatory Mechanisms*

Inga Sjuts, Jürgen Soll and Bettina Bölter

*61 Post-translational Modifications in Regulation of Chloroplast Function: Recent Advances*

Magda Grabsztunowicz, Minna M. Koskela and Paula Mulo


### **Chapter 3. Chloroplast-nucleus communication**


Yoshihiro Hirosawa, Yasuko Ito-Inaba and Takehito Inaba

# Editorial: Plastid Proteostasis: Relevance of Transcription, Translation, and Post-translational Modifications

#### Fiammetta Alagna<sup>1</sup> \*, Michele Bellucci <sup>2</sup> \*, Dario Leister <sup>3</sup> \* and Andrea Pompa<sup>2</sup> \*

<sup>1</sup> Viticoltura ed Enologia, Consiglio per la Ricerca in Agricoltura e l'Analisi dell'Economia Agraria (CREA), Turi, Italy, <sup>2</sup> Istituto di Bioscienze e Biorisorse, Consiglio Nazionale delle Ricerche (CNR), Perugia, Italy, <sup>3</sup> Plant Molecular Biology, Department Biology I, Ludwig-Maximilians-Universität, München, Germany

Keywords: plastome, gene expression, protein balance, regulation, nuclear-plastid interactions

### **Editorial on the Research Topic**

### **Plastid Proteostasis: Relevance of Transcription, Translation, and Post-translational Modifications**

Plastids are sites of biochemical and biological processes that are fundamental for plant life. The genome of these endosymbiotic organelles encodes for almost one hundred of the three thousand proteins that make up the chloroplast proteome. Genes coding for plastid multi-subunit protein complexes derive from both nuclear and plastid genomes, so it is clear that there is the need of a highly integrated coordination between this two subcellular compartments.

### Edited and reviewed by:

Julian Eaton-Rye, University of Otago, New Zealand

#### \*Correspondence:

Fiammetta Alagna fiammetta.alagna@crea.gov.it Michele Bellucci michele.bellucci@ibbr.cnr.it Dario Leister leister@lmu.de Andrea Pompa andrea.pompa@ibbr.cnr.it

#### Specialty section:

This article was submitted to Plant Cell Biology, a section of the journal Frontiers in Plant Science

Received: 11 August 2017 Accepted: 26 September 2017 Published: 17 October 2017

#### Citation:

Alagna F, Bellucci M, Leister D and Pompa A (2017) Editorial: Plastid Proteostasis: Relevance of Transcription, Translation, and Post-translational Modifications. Front. Plant Sci. 8:1759. doi: 10.3389/fpls.2017.01759

The coordination between the nucleus and the plastome takes place at many different levels, including the modulation of nuclear and plastid transcription, RNA processing and translation, post-translational modifications, and protein targeting. In addition, the plastid retains a whole series of mechanisms for the preservation of its protein balance (proteostasis), including proteases and molecular chaperones (**Figure 1**).

Plastids have largely abandoned transcriptional control switching predominantly to translational and post-translational control of their gene expression, but some transcriptional regulation is known to occur. Transcription of plastid genes is performed by two different types of RNA polymerases: plastid-encoded RNA polymerase (PEP) and a nuclear-encoded RNA polymerases (NEP). Liebers et al. propose that targeted changes in plastid transcription, mostly by controlling the relative activities of NEP and PEP enzymes, impact the establishment of the plastid proteome and these represent key determinants for the transitions between the different plastid types.

The transcriptional regulation mechanisms are still far from being completely elucidated. An unusual light- and stress-responsive promoter (psbD LRP), regulated by a AAG-box immediately upstream of the –35 element, has been recently mapped. Shimmura et al. analyzed psbD LRP promoter regions in 11 embryophytes, at different evolutionary stages, from liverworts to angiosperms. This analysis identified conserved features of the promoter and facilitated study of its emergence and evolution in plant species.

Among the proteins that regulate plastid gene expression, the nucleus-encoded proteins of the mitochondrial transcription termination factor (mTERF) family have been recently identified. Information on their function is only beginning to emerge. Xu et al. investigate the function of the chloroplast-associated mTERF. They report that these proteins are localized to chloroplast nucleoids and identify two of them involved in the salt stress response.

The import of plastid precursor proteins into plastids is another checkpoint affecting plastid proteostasis that is regulated in response to the fluctuating environmental conditions. This fine regulation ensures the optimal functioning of important biological processes

taking place in this cellular compartment. The import of plastid precursor proteins is mediated by two distinct translocation complexes called TOC and TIC, located respectively at the outer and at the inner envelope membrane of chloroplasts. The individual steps involved in protein translocation and the corresponding regulation mechanisms used by plants to modulate protein import are reviewed by Sjuts et al.

Upon transition from an endosymbiont to a plant cell organelle, the plastid retains a set of mechanisms that involves enzymes and proteins of prokaryotic origin which are responsible for protein maturation, post-translational modification, correct folding, protein abundance control, and removal of misfolded or damaged components. These processes ensure that plastid proteins are ready to exert their biological mission and require an intricate system of signals from nucleus to plastid and backwards. Plastid-derived signals can regulate availability of nuclear-encoded plastid precursors controlling their de novo synthesis, and targeting. Recently, an important player in the chloroplast-to-nucleus retrograde communication has been identified: the protein GENOMES UNCOUPLED1 (GUN1). Recent studies indicate that GUN1 might play a role in the coordination of translation, import, and degradation of plastid proteins. The molecular function of different GUN1 partners has been reviewed by Colombo et al., highlighting its potential role in plastid proteostasis. Another important mechanism during nuclear-plastid interaction is the degradation of multiple components through the ubiquitin–proteasome system. It has become increasingly clear that, together with feedback regulation of nuclear gene expression by plastid-derived signals, this mechanism avoids the accumulation of non-imported proteins in the cytosol. In addition to the anterograde signaling pathway, recent studies in A. thaliana demonstrated that also the retrograde signaling pathway can be subjected to ubiquitin– proteasome regulation. Hirosawa et al. review recent advances in understanding how the ubiquitin–proteasome system regulates the nuclear–plastid interaction and plastid biogenesis.

The maturation of plastid proteins is another highly regulated process that in some cases needs complex apparatuses to occur. This is the case of c-type cytochromes that require a multicomponent assembly pathway for their maturation, as reviewed by Gabilly and Hamel. The regulation of this pathway has not yet been clarified. Cline et al. describe the functional characterization of the CCS2 gene of Chlamydomonas reinhardtii required for cytochrome c assembly. The authors discuss the possible functions of CCS2 in the heme attachment reaction.

A wide range of post-translational modifications (PTMs) contribute to finely regulate the biological processes that take place in plastids. PTMs alter the physicochemical properties of the plastidial proteins thus affecting their function. Grabsztunowicz et al. review the current knowledge on the PTMs regulating important metabolic processes in chloroplasts such as DNA replication and gene expression, photosynthetic carbon assimilation, and starch metabolism and report the known physiological effects of these modifications.

Considering that there is limited knowledge of the combined action among the molecular mechanisms that regulate plastid proteostasis, the goal of this Research Topic is to bring together a set of articles that contribute to our understanding of how transcription, translation, and post-translational modifications (including protein targeting) maintain plastid proteostasis. As a consequence, this Research Topic will help us understand more deeply how plastids function.

### AUTHOR CONTRIBUTIONS

MB, DL, and AP prepared the research topic. All the authors edited the manuscripts. FA invited the authors and wrote the editorial. All the authors read and approved the editorial.

### ACKNOWLEDGMENTS

Research in the lab of Dario Leister is supported by the "Deutsche Forschungsgemeinschaft (DFG)" (grant number TRR175/C05).

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Alagna, Bellucci, Leister and Pompa. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Regulatory Shifts in Plastid Transcription Play a Key Role in Morphological Conversions of Plastids during Plant Development

Monique Liebers, Björn Grübler, Fabien Chevalier, Silva Lerbs-Mache, Livia Merendino, Robert Blanvillain and Thomas Pfannschmidt\*

Laboratoire de Physiologie Cellulaire et Végétale, Institut de Biosciences et Biotechnologies de Grenoble, CNRS, CEA, INRA, Université Grenoble Alpes, Grenoble, France

#### Edited by:

Fiammetta Alagna, The Council for Agricultural Research and Economics, Italy

#### Reviewed by:

Takashi Shiina, Kyoto Prefectural University, Japan Pal Maliga, Rutgers University, USA Yan Zubo, Dartmouth college, USA

#### \*Correspondence:

Thomas Pfannschmidt Thomas.Pfannschmidt@univgrenoble-alpes.fr

#### Specialty section:

This article was submitted to Plant Cell Biology, a section of the journal Frontiers in Plant Science

Received: 30 November 2016 Accepted: 05 January 2017 Published: 19 January 2017

#### Citation:

Liebers M, Grübler B, Chevalier F, Lerbs-Mache S, Merendino L, Blanvillain R and Pfannschmidt T (2017) Regulatory Shifts in Plastid Transcription Play a Key Role in Morphological Conversions of Plastids during Plant Development. Front. Plant Sci. 8:23. doi: 10.3389/fpls.2017.00023 Plastids display a high morphological and functional diversity. Starting from an undifferentiated small proplastid, these plant cell organelles can develop into four major forms: etioplasts in the dark, chloroplasts in green tissues, chromoplasts in colored flowers and fruits and amyloplasts in roots. The various forms are interconvertible into each other depending on tissue context and respective environmental condition. Research of the last two decades uncovered that each plastid type contains its own specific proteome that can be highly different from that of the other types. Composition of these proteomes largely defines the enzymatic functionality of the respective plastid. The vast majority of plastid proteins is encoded in the nucleus and must be imported from the cytosol. However, a subset of proteins of the photosynthetic and gene expression machineries are encoded on the plastid genome and are transcribed by a complex transcriptional apparatus consisting of phage-type nuclear-encoded RNA polymerases and a bacterial-type plastid-encoded RNA polymerase. Both types recognize specific sets of promoters and transcribe partly over-lapping as well as specific sets of genes. Here we summarize the current knowledge about the sequential activity of these plastid RNA polymerases and their relative activities in different types of plastids. Based on published plastid gene expression profiles we hypothesize that each conversion from one plastid type into another is either accompanied or even preceded by significant changes in plastid transcription suggesting that these changes represent important determinants of plastid morphology and protein composition and, hence, the plastid type.

Keywords: plastids, plastid morphology, photomorphogenesis, plant development, transcription, gene regulation, NEP, PEP

### INTRODUCTION

Plastids are cellular organelles that can be found only in plant and algae cells. They are of endosymbiotic origin that traces back to an evolutionary event in which a mitochondriate eukaryote took up a photosynthetically active cyanobacteria-like bacterium and established it as a permanent component of the cell, likely with the help of Chlamydiae (Ball et al., 2016). The

most prominent benefit for the eukaryotic cell in this process was the gain of photosynthesis and the concomitant switch from a heterotrophic to an autotrophic lifestyle (Hohmann-Marriott and Blankenship, 2011). The establishment of a stable endosymbiosis was, however, not an immediate evolutionary jump but a longongoing adaptation process in which the engulfed cyanobacterialike ancestor has lost slowly most of its genetic information toward the nucleus of the host cell by horizontal gene transfer (Abdallah et al., 2000; Martin et al., 2002; Reyes-Prieto et al., 2007). Only a small, but highly conserved set of genes finally remained encoded in the plastids' own genome of present plants, the plastome (Bock, 2007; Wicke et al., 2011). The vast majority of the proteome of present-day plant plastids is, therefore, encoded in the nucleus and must be imported from the cytosol (Rolland et al., 2012; Demarsy et al., 2014). Nevertheless, the proper expression of plastid genes is absolutely essential for the build-up of protein complexes involved in plastid gene transcription and translation as well as in metabolic processes such as photosynthesis or fatty acid biosynthesis (Jarvis and Lopez-Juez, 2013; Lyska et al., 2013). All major plastid multisubunit protein complexes are composed of a patchwork of nuclear and plastid encoded subunits and can be established only by a tight coordination of gene expression between the two genetic compartments (Pogson et al., 2015).

Alongside with these molecular and sub-cellular constraints, the establishment of plastid proteomes is strongly influenced by tissue-dependent and environmental cues. Multicellular, terrestrial plants are comprised of different organs with very divergent tissue organization and function. Plastids in these different tissues display large morphological and functional variations which are tightly connected to the function of the corresponding tissue (Schnepf, 1980; Lopez-Juez and Pyke, 2005). An individual plant, thus, possesses several different plastid types that represent distinct manifestations of the same cell organelle. Interestingly, most of these plastid types can interconvert upon environmentally induced changes in plant and tissue development. These morphological and functional conversions are only possible by corresponding changes in the plastid proteome composition. In this mini-review we focus on the specific changes in plastid gene expression that occur before or during transitions between different plastid types in the course of plant development.

### The Different Plastid Types of Plant Cells

Plant cells cannot generate plastids de novo but they gain them by inheritance from their progenitor cell. During division of the mother cell plastids are distributed arbitrarily between daughter cells and multiply afterward, by fission using a prokaryotictype division apparatus (Osteryoung and Pyke, 2014). The final number of plastids within a cell is cell-type specific and depends on regulatory mechanisms that are far from being understood yet (Cole, 2016). In addition, an individual cell does typically contain only one specific plastid type indicating that plastid development and cell development are interlinked. The various developmental lines and possible conversions between plastid types are subsequently discussed using the life cycle of the angiosperm Arabidopsis as a model (**Figure 1**). Because of space constraints detailed species-specific differences or special cases will be not considered here.

In Arabidopsis (like in most angiosperms) plastids are inherited maternally as a undifferentiated and small precursor form called proplastid (Pyke, 2007). In other species proplastids might be inherited also by paternal or biparental means. Knowledge in this field is poor and active regulation mechanisms remain to be clarified (Greiner et al., 2015). After fertilization of the egg cell Arabidopsis embryos undergo a morphological program typical for angiosperms that eventually ends with dry seeds (Le et al., 2010; compare **Figure 1**, outer circle). Like in many other oilseed crops Arabidopsis embryogenesis is characterized by an intermediate photosynthetically active period in which proplastids develop into chloroplasts in a stagespecific manner (Tejos et al., 2010). Chloroplast containing cells are already detected at the globular stage, but are most abundant during 6–12 days after fertilization (Allorent et al., 2013). This phase appears to be important for the fitness of the seed (Allorent et al., 2015). In a subsequent desiccation phase these chloroplasts then de-differentiate into non-photosynthetic, colorless leucoplasts, called eoplasts (Mansfield and Briarty, 1991, 1992). After seed imbibition and germination these eoplasts then re-differentiate into various plastid types depending on tissue context and environmental conditions.

In the dark, seedlings follow a developmental program called skotomorphogenesis (Solymosi and Schoefs, 2010). In cotyledons of such seedlings eoplasts develop into etioplasts while those located in hypocotyl and root develop into different types of colorless leucoplasts that are difficult to distinguish at the morphological level. Etioplasts are characteristic for this developmental program and represent an intermediate stand-by state of chloroplast formation. They do not develop a thylakoid membrane system, but a prolamellar body (PLB) that is composed of regular arrangements of NADPH, the enzyme protochlorophyllide-oxido-reductase (POR), the chlorophyll precursor protochlorophyllide and the thylakoid membrane lipids digalactosyl-diacylglycerol (DGDG) and monogalactosyldiacylglycerol (MGDG; Bastien et al., 2016). Upon illumination another developmental program called photomorphogenesis is initiated by the phytochrome-mediated photoreceptor network that triggers the expression of many nucleus located genes coding for chloroplast proteins (Arsovski et al., 2012). In parallel, thylakoid membranes begin to form and the lightdependent POR induces chlorophyll biosynthesis within the PLB. Chloroplast biogenesis then is usually completed after just 6–24 h. If seeds germinate directly in light the skotomorphogenic program is skipped and the eoplasts within the cotyledons differentiate directly into chloroplasts. Whether or not proplastids and eoplasts represent fully equivalent developmental stages remain to be elucidated. Studies on the transition of dedifferentiated desiccoplasts into etio- or chloroplasts after rehydration and illumination in the poikilochlorophyllous plant Xerophyta humilis may provide novel clues for the understanding of proplastid/eoplast-to-chloroplast transitions (Solymosi et al., 2013).

During primary leaf formation chloroplasts originate directly from proplastids present in the shoot apical meristem (SAM;

(indicated by rolling lines). Changes in plastid transcriptional apparatus or activity that occur during these transitions are indicated by symbols NEP and PEP. Size of letters represents the relative activities of the two types of RNA polymerases in the respective plastid type. For details see text.

Charuvi et al., 2012). Fully developed chloroplasts in green parts of plants multiply then by fission until they reach the celltype specific number. This, however, accounts mainly for the mesophyll tissue while in the epidermis a likely tissue-specific program leads to a differential development of chloroplasts. Guard cells were reported to display high numbers of fully developed chloroplasts while pavement cells contain rather low numbers of relatively small chloroplasts (around half the size of those in mesophyll cells) that may contain reduced levels of chlorophyll (Barton et al., 2016). In reproductive organs such as fruits or flowers chloroplasts usually transform into chromoplasts as part of maturation or developmental programs. In senescing tissues the valuable resources of chloroplasts, notably the nitrogen bound in chlorophylls and photosynthesis proteins such as RubisCO, are reallocated and the plastids turn into gerontoplasts, the aging form of plastids.

In hypocotyls and roots of growing seedlings eoplasts usually develop into a number of colorless plastids commonly summarized under the term leucoplasts. This group comprises amyloplasts, statoliths, and elaioplasts (**Figure 1** and **Table 1**) and, in later stages, may develop also in other parts of the plant. These colorless plastids do develop even if the tissues are exposed to light. This strongly suggests that the transition from proplastids/eoplasts into chloroplasts is actively inhibited in these tissues, likely by internal factors. Recent studies have demonstrated that the developmental block of chloroplast development in Arabidopsis roots can be released either genetically or by external hormone treatment (Kobayashi et al., 2012) supporting the view of an active inhibition in chloroplast biogenesis in these tissues. Release of such an inhibition represents not only an artificial effect but does occur also under physiological conditions as some studies reported the presence of fully developed chloroplasts in Arabidopsis hypocotyls (Jin et al., 2001; Hermkes et al., 2011). These chloroplasts were found to be involved in phototropic responses suggesting that they play a defined physiological role (Jin et al., 2001). Studying the mechanisms that control this eoplastchloroplast transition could help to understand principle steps


#### TABLE 1 | Summary of major plastid types in plant cells.

fpls-08-00023 January 17, 2017 Time: 16:47 # 4

Typical plastid types found in vascular plants are listed. Leucoplasts are not included as they represent a group of plastids (summarizing all non-green plastids lacking pigments including amyloplasts) rather than defining a specific plastid type. The group of leucoplasts contains also other less prominent plastid types that are not well investigated and not discussed in this review such as root plastids and proteinoplasts. For more information on these specialized plastid forms readers are referred to corresponding reviews (Schnepf, 1980; Pyke, 2007).

of early chloroplast biogenesis and to identify novel regulatory factors of plastid transitions (Chiang et al., 2012).

### Shifts in Plastid Transcription during Morphological Transitions of Plastids

The different types of plastids mentioned above perform very different functions that are highly specific for the tissue in which they reside (Pyke, 2007) (**Table 1**). Despite their morphological and functional diversity they all contain the same genome (Bock, 2007). However, their strong functional diversity implies a specific enzymatic configuration for each plastid type. This requires a controlled adjustment in the expression of both plastid and nuclear genes encoding the proteins for each of these specific organelle manifestations. Here, we focus on the adjustment of plastid gene expression.

Molecular and genetic studies uncovered that transcription of plastid genes is performed by two different types of RNA polymerases. One type is comprised by two single-subunit phage-type RNA polymerases encoded by two different nuclear genes (nuclear-encoded RNA polymerases, NEP). These proteins are targeted either only to the plastid (RpoTp) or dually to plastids and mitochondria (RpoTmp). The other type of RNA polymerase is a multi-subunit enzyme of prokaryotic type with four basic subunits encoded in the plastid genome (RpoA, RpoB, RpoC1, and RpoC2; plastid-encoded RNA polymerase, PEP). For promoter recognition this enzyme complex is dependent on the interaction with sigma factors (called Sig1 – Sig6 in Arabidopsis) that are encoded in the nucleus (Toyoshima et al., 2005; Schweer et al., 2010; Lerbs-Mache, 2011; Borner et al., 2015; Pfannschmidt et al., 2015). The two types of RNA polymerase activities utilize different promoters and depending on their respective promoter structure the genes on the plastid genome can be categorized into three different classes. Class I comprises genes possessing only PEP promoters (only photosynthesis genes). Class II covers genes that have both NEP and PEP promoters (most other genes including genes for the ATP synthase and many components of the gene expression system). Class III represents genes with NEP promoters only and comprises ycf2 (encoding a still unknown protein), accD (encoding the β-carboxyltransferase subunit of the acetyl CoA carboxylase) and the rpoBC1C<sup>2</sup> operon (Liere et al., 2011). This diversity of promoter structures and the multiplicity of transcriptional components (see also below) represent a prerequisite for efficient transcriptional regulation during plastid conversion where plastid housekeeping genes are preferentially transcribed by NEP and photosynthesis related genes are transcribed by PEP (Allison et al., 1996; Hajdukiewicz et al., 1997).

We propose that targeted changes in plastid transcription, mostly by controlling the relative activities of NEP and PEP enzymes, impact the establishment of the plastid proteome and, therefore, represent key determinants for the transitions between the different plastid types.

#### Proplastid/Eoplast-Chloroplast Transition

Proplastids can be found only in meristematic cells of plants and in in vitro cultured cells. Isolation of proplastids from meristematic cells is technically not feasible. However, as meristematic cells give rise to various plant organs, proplastids might be considered as starting point for differentiationdependent plastid conversion. Also, plastid gene expression in proplastids and after controlled conversion of proplastids into amyloplasts has been analyzed using in vitro cultured cells [(Sakai et al., 1992), see below]. These early experiments already showed that such plastid conversion is accompanied by changes in plastid transcriptional activity.

Proplastid/eoplast-chloroplast conversion-associated changes in plastid gene expression patterns have been characterized

in detail during Arabidopsis seed formation and germination (Demarsy et al., 2012; Allorent et al., 2013). Although slight increases of NEP transcribed mRNAs were observed in this transition, the predominant changes concern remarkable increases of mRNAs of photosynthesis related proteins. If proplastid/eoplast-chloroplast conversion is prevented by deletion of plastid rpo genes, colorless plastids of 2–5 µm length are formed that might be considered as a genetically induced type of leucoplasts (Allison et al., 1996; De Santis-MacIossek et al., 1999). Thus, establishment of the correct NEP/PEP configuration and their relative activities at a given developmental stage is absolutely essential for successful chloroplast differentiation.

### Etioplast-Chloroplast Transition

Etioplasts and their light-induced transition to chloroplasts are well studied in numerous dicotyledonous and monocotyledonous species. Most striking is the very rapid development of thylakoid membranes, increase in chlorophyll content and construction of the photosynthetic apparatus that requires both a massive import of nuclear encoded plastid proteins and high expression of plastid-encoded genes (Lonosky et al., 2004; von Zychlinski et al., 2005; Philippar et al., 2007; Pudelski et al., 2009; Majeran et al., 2010; Ploscher et al., 2011). Etioplasts display just a basic transcriptional activity and accumulate photosynthesis transcripts only to very low levels. Shifting dark-grown seedlings to light, however, rapidly induce a plastome-wide transcript accumulation of photosynthesis genes reaching a maximum level after 10–44 h mRNA levels followed by decrease to approximate pre-illumination levels (Rodermel and Bogorad, 1985). The initial increase in mRNA is followed by subsequent translation of the corresponding proteins (Kanervo et al., 2008). It should be noted that tissue-specific gene expression analyses distinguishing epidermal and mesophyll tissues were never reported and that the results in all studies to date, thus, represent a mixture of both cell types. This is critical with respect to the notion that recent studies suggest a specific sensor function for epidermal chloroplasts (Virdi et al., 2015, 2016). Targeted research on this special type of chloroplasts will be required in order to understand their detailed physiological function.

The light-dependent activation of plastid gene expression during etioplast-chloroplast conversion includes posttranslational modifications such as phosphorylation of PEP subunits and sigma factors (Tiller and Link, 1993) and a restructuring of the PEP complex. While in etiolated mustard seedlings PEP was found to exist in its prokaryotic composition (α2, β, β 0 , β <sup>00</sup> subunits), a much larger PEP complex with many additional subunits was purified from fully developed chloroplasts. Studies on intermediate plastids isolated from seedlings illuminated for just 16 h identified both complexes to around equal activities suggesting a lightinduced conversion between these two plastid RNA polymerase complexes (Pfannschmidt and Link, 1994). Detailed mass spectrometry analyses identified these subunits and a set of conserved PEP-associated proteins (PAPs) could be defined (Pfannschmidt et al., 2000; Suzuki et al., 2004; Pfalz et al., 2006; Steiner et al., 2011).

PEP-associated proteins are all nuclear encoded and are rapidly light-induced during etioplast-chloroplast transition yielding the observed PEP restructuring (Yagi et al., 2012). Genetic inactivation of any of these PAPs in Arabidopsis, maize or rice results in a block of proper chloroplast development and ends up in albinoic phenotypes suggesting that pap gene expression and/or subsequent PEP re-structuring represent essential steps in early chloroplast biogenesis. Evolutionary presence of pap genes appears to be restricted to terrestrial plants and ferns suggesting that their appearance is connected to the conquest of land (Pfalz and Pfannschmidt, 2013). These genes, thus, likely represent an evolutionary indicator for the development of chloroplast-containing multi-cellular plants (de Vries et al., 2016).

Plastid gene expression changes during etioplast to chloroplast conversion were also analyzed in the monocotyledonous plant maize. In monocotyledons, leaf development is initiated at a basal meristem resulting in a gradient of chloroplast development from the bottom to the tip (Baumgartner et al., 1989, 1993; Hess et al., 1993). This gradient has been used extensively as a model for chloroplast biogenesis. About 51 plastid genes were found to be at least two times higher expressed in tips than in the leaf base (Cahoon et al., 2008). It is, however, still debated how far this plastid developmental gradient reflects the corresponding situation (proplastid-to-chloroplast conversion) in dicotyledonous plants.

### Chloroplast-Chromoplast Transition

Chromoplasts mainly develop from chloroplasts in formerly green plant tissues e.g., during fruit ripening or flower development. They can also develop directly from proplastids or amyloplast depending on species and tissue (Egea et al., 2010). Plastid gene expression during conversion of green chloroplasts toward red chromoplasts has been characterized in detail during tomato fruit ripening. In contrast to the rapid etioplast-chloroplast transition in cotyledons, the chloroplastchromoplast transition in tomato fruits requires several days or even weeks allowing transcript analyses of various intermediary stages. These studies uncovered both systemic and gene-specific effects (Kahlau and Bock, 2008). Most important, green tomato fruits displayed a dramatic reduction in chloroplast transcripts compared to green leaves from the same plant. This indicates that the fruit developmental program provides a dominant repressive impact on plastid transcriptional activities even before ripening effects became visible. This may prevent the unnecessary production of photosynthesis proteins in the tomato fruit already in early stages of ripening.

In contrast, changes in plastid gene expression in the subsequent stages (turning, light red, red) remained relatively subtle suggesting that the chloroplast-chromoplast conversion itself is not accompanied by major changes in plastid gene expression. An exception was observed for the accD gene that displayed a targeted accumulation at both, transcript and protein levels. Accumulation of the protein AccD as part of the fatty acid biosynthesis complex may allow the accumulation of lipids necessary for storage of carotenoids produced during fruit ripening. AccD gene expression requires at least a low level of expression of genes involved in transcription/translation. Indeed, the repression by the fruit developmental program was stronger for photosynthesis genes than for genetic system genes (Kahlau and Bock, 2008) suggesting that low levels of plastid gene expression activity may remain. These remaining activities may be directed to the observed targeted accD gene expression.

#### Proplastid-Amyloplast Transition

fpls-08-00023 January 17, 2017 Time: 16:47 # 6

Amyloplasts are the plastids of storage organ tissues and roots and typically contain high amounts of starch. Systematic gene expression studies in this plastid type were done using potato tubers (Brosch et al., 2007; Valkov et al., 2009). When compared to leaf chloroplasts tuber amyloplasts displayed very low levels of gene expression in terms of transcriptional rate, transcript accumulation and maturation as well as ribosome association of mRNAs and translation. Both, NEP and PEP enzymes are present, but run-on transcription experiments revealed very low transcriptional rates of both enzyme activities. Interestingly, like in chromoplasts accD expression appeared to be an exception. It displayed relatively stable transcript levels and ribosome association (Valkov et al., 2009) thus confirming the importance of AccD for the maintenance of plastids regardless of their morphological type. In addition, trans-plastomic inactivation of the plastid accD gene in tobacco revealed to be impossible (Kode et al., 2005).

Tissue cultures of tobacco bright-yellow (BY)-2 cells represent another test system to study amyloplasts (Miyazawa et al., 1999; Enami et al., 2011). In presence of cytokinin these cells develop amyloplasts from proplastids. Microarray analysis of the transcriptome did not reveal specific changes between the two plastid types, including the accD gene. Interestingly, inhibitors of plastid transcription or translation blocked the hormone-induced differentiation of amyloplasts indicating signaling of plastid gene expression to the hormone-induced plastid developmental pathway. This specific retrograde signaling pathway seems to act via intermediates of tetrapyrrole biosynthesis, i.e., haem (Enami et al., 2011).

### REFERENCES


### CONCLUSION AND PERSPECTIVES

In all plastid conversions investigated so far, changes in the plastid transcriptional apparatus and/or transcriptional activity either accompany or even precede the transition. Proper control of plastome transcription, thus, appears to be an important determinant for these developmental steps. Future research will focus on the identification of regulators that may serve as master switches of plastid development in response to internal and external cues (Lopez-Juez, 2007). In addition, more detailed studies on gene expression in proplastids or eoplasts may be highly informative for understanding the molecular regulation of plastid development especially in their initial steps.

### AUTHOR CONTRIBUTIONS

ML contributed a figure, contributed to the manuscript text, read, and approved the final version. BG contributed to the manuscript text, read, and approved the final version. FC contributed to the manuscript text, read, and approved the final version. SL-M contributed to the manuscript text, read, and approved the final version. LM contributed to the manuscript text, read, and approved the final version. RB contributed to the manuscript text, read, and approved the final version. TP developed manuscript idea, wrote the manuscript with the help of all co-authors.

### FUNDING

Work in our team was supported by grants from the Deutsche Forschungsgemeinschaft to TP (PF323-4-2; PF323-5-2) and the DFG research group FOR 804. Our studies received also institutional financial support from the French National Research Agency (ANR-10-LABEX-04 GRAL Labex, Grenoble Alliance for Integrated Structural Cell Biology), the University Grenoble-Alpes (AGIR program to RB) and the CNRS (PEPS-ExoMod program to TP).




**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Liebers, Grübler, Chevalier, Lerbs-Mache, Merendino, Blanvillain and Pfannschmidt. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Comparative Analysis of Chloroplast psbD Promoters in Terrestrial Plants

Shuichi Shimmura<sup>1</sup>† , Mikio Nozoe1,2† , Shota Kitora<sup>1</sup> , Satoko Kin<sup>1</sup> , Shigeru Matsutani1,3 , Yoko Ishizaki<sup>1</sup> , Yoichi Nakahira1,4 and Takashi Shiina<sup>1</sup> \*

<sup>1</sup> Graduate School of Life and Environmental Sciences, Kyoto Prefectural University, Kyoto, Japan, <sup>2</sup> AMITA Institute for Sustainable Economies Co., Ltd., Kyoto, Japan, <sup>3</sup> Kyoto Botanical Garden, Kyoto, Japan, <sup>4</sup> College of Agriculture, Ibaraki University, Ibaraki, Japan

The transcription of photosynthesis genes encoded by the plastid genome is mainly mediated by a prokaryotic-type RNA polymerase called plastid-encoded plastid RNA polymerase (PEP). Standard PEP-dependent promoters resemble bacterial sigma-70-type promoters containing the so-called −10 and −35 elements. On the other hand, an unusual light- and stress-responsive promoter (psbD LRP) that is regulated by a 19-bp AAG-box immediately upstream of the −35 element has been mapped upstream of the psbD-psbC operon in some angiosperms. However, the occurrence of the AAG-box containing psbD LRP in plant evolution remains elusive. We have mapped the psbD promoters in eleven embryophytes at different evolutionary stages from liverworts to angiosperms. The psbD promoters were mostly mapped around 500–900 bp upstream of the psbD translational start sites, indicating that the psbD mRNAs have unusually long 5<sup>0</sup> -UTR extensions in common. The −10 elements of the psbD promoter are well-conserved in all embryophytes, but not the −35 elements. We found that the AAG-box sequences are highly conserved in angiosperms and gymnosperms except for gnetaceae plants. Furthermore, partial AAG-box-like sequences have been identified in the psbD promoters of some basal embryophytes such as moss, hornwort, and lycophyte, whereas liverwort has the standard PEP promoter without the AAG-box. These results suggest that the AAG-box sequences of the psbD LRP may have evolved from a primitive type of AAG-box of basal embryophytes. On the other hand, monilophytes (ferns) use another type of psbD promoter composed of a distinct cis-element upstream of the potential −35 element. Furthermore, we found that psbD expression is not regulated by light in gymnosperms or basal angiosperms, although they have the well-conserved AAG-box sequences. Thus, it is unlikely that acquisition of the AAG-box containing psbD promoter is directly associated with light-induced transcription of the psbD-psbC operon. Light- and stress-induced transcription may have evolved independently and multiple times during terrestrial plant evolution.

Keywords: psbD LRP, chloroplast, promoter, evolution, stress

### INTRODUCTION

Chloroplasts in plant and algal cells are semiautonomous organelles that have their own genome and gene expression system, reflecting their cyanobacterial origin. Chloroplast transcription is mediated by two distinct RNA polymerase systems, a prokaryotic multi-subunit RNA polymerase (PEP) whose core subunits are encoded by chloroplast genomes and single-subunit

#### Edited by:

Dario Leister, Ludwig-Maximilians-Universität München, Germany

#### Reviewed by:

Karin Krupinska, University of Kiel, Germany Kristina Kuehn, Humboldt University of Berlin, Germany

\*Correspondence: Takashi Shiina shiina@kpu.ac.jp †These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Plant Cell Biology, a section of the journal Frontiers in Plant Science

Received: 31 March 2017 Accepted: 21 June 2017 Published: 13 July 2017

#### Citation:

Shimmura S, Nozoe M, Kitora S, Kin S, Matsutani S, Ishizaki Y, Nakahira Y and Shiina T (2017) Comparative Analysis of Chloroplast psbD Promoters in Terrestrial Plants. Front. Plant Sci. 8:1186. doi: 10.3389/fpls.2017.01186

bacteriophage-type RNA polymerases (NEP) that are encoded by the nuclear genome (Hess and Börner, 1999; Liere et al., 2011; Yagi and Shiina, 2012, 2014; Liebers et al., 2017). The PEP core enzyme consists of four major subunits, designated as α, β, β 0 , and β <sup>00</sup>" subunits, which are homologous to bacterial subunits. Another dissociable subunit called a sigma factor allows the core enzyme to initiate transcription from the specific promoters. Multiple sigma factor genes have been identified in embryophytes (Tanaka et al., 1997; Morikawa et al., 1999; Fujiwara et al., 2000; Hara et al., 2001; Kasai et al., 2004; Kubota et al., 2007; Kanazawa et al., 2013), and they play specific roles in transcriptional regulation in response to developmental and/or environmental cues (reviewed by Kanamaru and Tanaka, 2004; Shiina et al., 2005; Schweer et al., 2010; Börner et al., 2015; Chi et al., 2015). Standard PEP-dependent promoters resemble bacterial sigma-70 type promoters containing −10 (TATAAT) and −35 (TTGACA) elements, reflecting their bacterial origin. NEP recognizes distinct types of promoters containing a core YRTA motif (Hess and Börner, 1999; Liere and Maliga, 1999; Liere et al., 2004; Börner et al., 2015). PEP transcribes mainly photosynthesis genes in mature chloroplasts while NEP transcribes housekeeping genes in both chloroplasts and non-photosynthetic plastids (Allison et al., 1996; Hajdukiewicz et al., 1997). The functional coordination of PEP and NEP plays a critical role in plastid differentiation in angiosperms.

In contrast to angiosperms, chloroplasts of the green algae C. reinhardtii harbor a simple transcription system, which is dependent on PEP and a single sigma factor SIG1 (Bohne et al., 2006; Yagi and Shiina, 2014). No NEP has been identified in Chlamydomonas. It is considered that embryophytes have developed complex transcription systems to adapt to marked environmental stresses. However, the evolutionary process of chloroplast transcription systems in embryophytes remains largely elusive.

Most PEP-dependent genes are actively transcribed in green tissues including the leaves. The chloroplast run-on experiments demonstrated that PEP-dependent transcription is activated by high light compared to normal growth light (Baena-González et al., 2001). However, the accumulation of most PEP-dependent transcripts is not regulated by light/dark transitions or environmental stresses, possibly due to the extraordinary stability of their transcripts (Shiina et al., 1998; Hayes et al., 1999). The only exception is a psbD light-responsive promoter designated psbD LRP, which is located upstream of a psbD-psbC operon encoding D2 (PsbD) and CP47 (PsbC) subunits of the PSII reaction center complex (Christopher et al., 1992; Wada et al., 1994; Allison and Maliga, 1995; To et al., 1996; Hoffer and Christopher, 1997). Transcription from the psbD LRP is activated by not only high-irradiance light, but also various abiotic stresses, including salt, high osmolarity and heat (Nagashima et al., 2004) and circadian rhythm (Noordally et al., 2013). The psbD LRP contains unique signature sequences named the AAG-box, immediately upstream of the -35 element (Allison and Maliga, 1995; Kim and Mullet, 1995; To et al., 1996; Nakahira et al., 1998; Kim et al., 1999). The AAG-box is composed of two different repeat units (AAGT and GACC/T repeats). In vitro transcription assays from the psbD LRP revealed that both repeat motifs are important for transcription, but not the −35 element in barley or wheat (Nakahira et al., 1998; Kim et al., 1999). Furthermore, the AAGT repeat interacts with the sequence-specific DNA-binding protein AGF (Kim and Mullet, 1995; Nakahira et al., 1998; Kim et al., 1999). It has also been shown that the stress-responsive plastid sigma factor SIG5 directs the activation of the psbD LRP in Arabidopsis thaliana (Nagashima et al., 2004; Tsunoyama et al., 2004; Onda et al., 2008).

The psbD promoter mapped in Chlamydomonas has a well-conserved −10 element, but lacks the AAG-box and standard −35 element (Klein et al., 1992; Klinkert et al., 2005). In addition, nucleotide sequence comparison of the upstream regions of the psbD among embryophytes suggests that A. thaliana (angiosperm) and Pinus thunbergii (gymnosperm) have the psbD LRP in their genome, but not the other basal embryophytes Physcomitrella patens (moss) and Marchantia polymorpha (liverwort) (Kanazawa et al., 2013). These findings suggest that the psbD LRP may have emerged during the evolution of embryophytes. However, evolution of the psbD promoter remains elusive. In this study, we mapped the promoter region of the psbD-psbC operon in eleven embryophytes at different evolutionary stages from liverwort to angiosperm. The results suggest that AAG-box sequences of the psbD LRP in angiosperms and gymnosperms may have evolved from the partial AAG-box-like sequences detected in the psbD promoters of basic embryophytes such as moss, hornwort, and lycophyte, while monilophytes (ferns) use a distinct type of psbD promoter lacking the AAG-box. On the other hand, lightdependent psbD expression was not observed in gymnosperms or primitive angiosperms that possess the well-conserved AAGbox, suggesting that the AAG-box containing psbD promoter acquisition is unlikely to be associated with the occurrence of light-dependent psbD expression.

### MATERIALS AND METHODS

### Plant Materials and Growth Condition

For primer extension analysis, A. thaliana, Adiantum capillusveneris, P, patens, and M. polymorpha (Tak-1) were grown in the light in growth chambers at 22◦C under 16-h-light/8-hdark (80–100 µmol photons m−<sup>2</sup> s −1 ). Other samples (Laurus nobilis, Ginkgo biloba, P. thunbergii, Equisetum hyemale, Psilotum nudum, and Lycopodium clavatum were collected from plants cultivated at Kyoto Botanical Garden. Leaf samples were collected in the daytime, and immediately frozen in liquid N2. Light-induced gene expression analysis was carried out with plants (A. capillus-veneris, C. revoluta, P. thunbergii, L. nobilis, A. thaliana) grown in the growth chambers at 22◦C under 16 h-light/8-h-dark (80 µmol photons m−<sup>2</sup> s −1 ). Plants were darkadapted for 72 h, and then exposed to light of 180 µmol photons m−<sup>2</sup> s −1 for 4 h in the growth chamber. Collected leaf samples were immediately frozen in liquid N2.

Light-induced gene expression analysis (**Supplementary Figures S4–S8**) was also carried out with a wide range of embryophytes at different evolutionary stages from moss to

angiosperms grown in the growth chambers at 22◦C under 16 h-light/8-h-dark (80 µmol photons m−<sup>2</sup> s −1 ). Plants were darkadapted for 72 h (D), and then illuminated (275 µmolm−<sup>2</sup> s −1 ) for up to 12 h (L). Osmotic stress was achieved by 250 mM mannitol treatment for 6 h to the detached leaves. Collected leaf samples were immediately frozen in liquid N2.

Arabidopsis thaliana wild-type Columbia ecotype and AtSIG5 overexpressing plants were germinated and grown on two layers of filter paper on the one-half Murashige and Skoog (MS) medium containing 0.8% (w/v) agar at 22◦C with 16-h light (80 µmol photons m−<sup>2</sup> s −1 )/8-h dark cycles for 10 days. For salt and high osmotic stress treatments, the seedlings were transferred to one-half MS medium containing 250 mM NaCl or 250 mM mannitol, and incubated under light (80 µmol photons m−<sup>2</sup> s −1 ) for 6–24 h. For low temperature treatment, the seedlings were incubated at 4◦C for 6–24 h under light conditions of 50 µmol photons m−<sup>2</sup> s −1 . For light response experiments, the seedlings were dark adapted for 72 h (D) and illuminated for 4 h with white light (80 µmol photons m–2 s–1) or a blue LED light (50 µmol photons m−<sup>2</sup> s −1 ) or a red LED light (50 µmol photons m−<sup>2</sup> s −1 ).

### Transgenic Plants

First strand cDNA of AtSIG5 was synthesized from total RNA prepared from Arabidopsis seedlings using AMV reverse transcriptase (TaKaRa). cDNA was amplified by PCR using KOD-plus-DNA polymerase (TOYOBO) according to the manufacturer's protocols. To obtain an AtSIG5 overexpression construct under the control of CaMV 35S promoter, the GUS gene of the binary vector pBI121 was replaced with the AtSIG5 cDNA. The resulting constructs were introduced into Agrobacterium tumefaciens and used to transform the wild-type (Col-0) plants.

### Total RNA Isolation, Primer Extension Analysis, and Northern Blot Analysis

Total RNA was extracted from the leaves using the RNeasy Plant Mini kit (Qiagen, United States) or TRIZOL <sup>R</sup> following the manufacturer's instructions. The primer extension assays were performed on the total RNA using the Primer Extension System (Promega) with the AMV reverse transcriptase following the manufacturer's instructions. Primers used are listed in Supplementary Table S1. Primer extension products were analyzed on a 6% polyacrylamide-7 M urea sequencing gel. For northern blot analysis, total RNA samples (2 µg) were separated by denaturing agarose gelelectrophoresis. After capillary blotting onto Hybond-N nylon membrane, RNA gel blots were hybridized to the randomly primed DNA probes for psbA and psbD of each plant. The psbD UTR probe (−1085 to −726 of the psbD translation start codon) was designed to detect specifically the transcripts from the psbD LRP in Arabidopsis. The psbD and psbA coding region probes were designed to detect transcripts produced from all multiple promoters in the psbD-psbC operon. The AtSIG5 probe was also prepared using specific primers. The psbD UTR probes specific for each plant were also generated by using PCR primers.

### RESULTS

### psbD LRP Transcription Is Dependent on SIG5

The psbD LRP is a unique PEP-dependent chloroplast promoter, which is responsible for the transcription of the psbD-psbC operon. The psbD-psbC operon is well-conserved among plants and cyanobacteria. Unlike standard PEP-dependent promoters composed of sigma-70 type −10 (TATAAT) and −35 (TTGACA) elements, it has been shown that psbD LRP activity is dependent on the upstream AAG-box in tobacco (Allison and Maliga, 1995), barley (Kim and Mullet, 1995), rice (To et al., 1996), and wheat (Nakahira et al., 1998; **Figure 1A**). On the other hand, the −35 element of the psbD LRP is not essential for transcription activity (To et al., 1996; Nakahira et al., 1998; Kim et al., 1999; Thum et al., 2001b). These findings suggest that the upstream AAG-box may take over the role of the pseudo −35 element in the psbD LRP.

Multiple promoters have been identified in the psbD-psbC operon (Hoffer and Christopher, 1997). The most upstream promoter (∼950 of the psbD translation start site) is a so-called psbD LRP that is specifically activated by blue light. To identify specifically mRNAs transcribed from the psbD LRP, we used a psbD UTR probe (−1085 to −726 of the psbD translation start codon) that is designed to be located upstream of the second promoter at −550 (Tsunoyama et al., 2004).

As shown in **Figure 1B**, the 4.5- and 3.7-kb transcripts from the psbD LRP were specifically detected by the psbD LRP UTR probe. As reported by Nagashima et al. (2004), various abiotic stresses including salt, cold, and hyperosmotic stresses induce transcription at the psbD LRP in a time-dependent manner (**Figure 1B**). Similarly, SIG5 transcription is activated by abiotic stresses. Previous reports (Nagashima et al., 2004; Tsunoyama et al., 2004) demonstrated that psbD LRP activity is abolished in AtSIG5-deficient mutants in Arabidopsis. In order to further define the role of AtSIG5 in transcription at the psbD LRP, we developed SIG5 overexpression lines (SIG5oxA and SIG5oxH) and examined transcription activity from the psbD LRP. The accumulation of psbD LRP transcripts was significantly increased by the overexpression of AtSIG5 in illuminated plants irrespective of the presence of white, blue, and red light, but only slightly in the dark (**Figure 1C**). These results clearly demonstrate that SIG5 specifically mediates transcription from the psbD LRP in the light. Photoreceptors including CRY1, CRY2, and PhyA have been shown to be involved in the light-induced expression of the psbD-psbC operon in Arabidopsis (Thum et al., 2001a), while AtSIG5 overexpression cannot activate transcription from the psbD LRP in the dark (**Figure 1C**). Taken together, photoreceptor-mediated signaling may modify SIG5 or SIG5 import in a light-dependent manner and activate transcription at the psbD LRP.

#### psbD Transcripts have Markedly Long 5 0 -UTRs in Common

Transcription initiation sites of the psbD-psbC operon have only been identified in some angiosperm model plants, including barley, wheat, rice, tobacco, and Arabidopsis (Yao et al., 1989;

Christopher et al., 1992; Wada et al., 1994; To et al., 1996; Hoffer and Christopher, 1997), and green algae Chlamydomonas reinhardtii (Klein et al., 1992). In order to address the evolutionary changes of psbD promoter structures, we mapped 5 0 -ends of psbD transcripts of eleven embryophytes at different evolutionary stages from liverworts to angiosperms using primer extension analysis (**Figure 2** and **Supplementary Figure S1**, **S2**). Leaf samples were collected from plants grown in Kyoto Botanical Garden in the daytime, except for A. thaliana, A. capillus-veneris, P. patens, and M. polymorpha that were grown in the light in growth chambers. We estimated the size of primer extension products approximately by comparing their mobility profiles to single-strand DNA ladders. In order to determine the start sites as exactly as possible, we designed appropriate primers that produce primer extension products shorter than 300 bases. Next we searched for sequences similar to the conserved −10 sequences (TATTCT) of the psbD LRP in close proximity to the identified transcription initiation sites. Then, we aligned the deduced psbD promoter sequences with those of other plants using the −10 element as reference. In this study, we considered the psbD transcripts with the most upstream terminus as the primary transcripts, except for P. thunbergii and E. hyemale, which have another promoter upstream of the potential psbD LRP.

Transcription initiation sites from the psbD LRP have been mapped at 572, 610, 566, 905, and 948 bp upstream of the psbD translation start site of barley (Christopher et al., 1992), wheat (Wada et al., 1994), rice (To et al., 1996), tobacco (Yao et al., 1989), and Arabidopsis (Hoffer and Christopher, 1997), respectively. Similarly, 5<sup>0</sup> -ends of the psbD primary transcripts of the psbD-psbC operon were mapped at 800–900 bp upstream of the psbD gene in the most angiosperms including the basal angiosperm L. nobilis, gymnosperm P. thunbergii, primitive gymnosperm G. biloba, monilophyte (Leptosporangiate fern) A. capillus-veneris, monilophyte (Eusporangiate fern) P. nudum, monilophyte (Eusporangiate fern) E. hyemale, and lycophyte Huperzia lucidula (**Figure 3**). Furthermore, we found sequences similar to the psbD promoter of H. lucidula at the far upstream position (−919) of the psbD translation start site in Anthoceros formosae (hornwort). On the other hand, 5<sup>0</sup> -ends of the longest psbD transcripts of other bryophytes, P. patens (moss), and M. polymorpha (liverwort) are located at −246 and −243 of the psbD gene, respectively. These results indicate that psbD mRNAs have unusually long 5<sup>0</sup> -UTR extensions in common, except for mosses and liverworts. It is of note that intergenic distances between psbD and the upstream trnT are much shorter in mosses and liverworts compared with those of other plants.

### AAG-Box of the psbD LRP Is Highly Conserved among Angiosperms and Gymnosperms, and Partially Conserved in Lycophytes and Mosses

Next, we compared sequences immediately upstream of the psbD transcription initiation sites that were mapped in this and previous studies. The typical −10 elements (TATTCT) are wellconserved in all psbD promoters (**Figure 4** and Supplementary Table S2). Conversely, the potential −35 elements are less conserved among terrestrial plants and show weak similarity (less than ∼50%) to the consensus sequences (TTGACA). On the other hand, liverworts possess a typical sigma-70 type promoter with conserved −35 and −10 elements with 18-nt spacing. As expected, the AAG-box is well-conserved among angiosperms and gymnosperms. The consensus sequence of the AAG-box is "RAAGTAAGTRRACCTRACYY," which contains an AAGT

are shown as numbers in parentheses. Lane C shows an experiment with the control RNA and primer provided by the manufacturer that produces an 87-base

repeat and a GACC/T repeat. The AAG-box sequences are almost 80% identical in most gymnosperms and angiosperms, including the primitive gymnosperm G. biloba and basal angiosperm

L. nobilis (Supplementary Table S2). We found that a 13 bp core sequence of the AAG-box is also highly conserved (∼85%) in lycophytes and hornworts, and partially conserved

primer extension product.

arrows. The psbD LRP-related promoters analyzed in this study are indicated by red shadows.

in mosses (69%), but not in liverworts. The AAG-box of lycophytes and hornworts harbors the GACC/T repeat-like sequences, but lacks the AAGT repeat (**Figure 4B**). On the other hand, neither the AAGT repeat nor GACC/T repeat are conserved in the psbD promoter of monilophytes (ferns). Instead, sequences upstream of the −35 element are wellconserved among standard monilophytes and P. nudum, but not in the primitive monilophyte E. hyemale (**Figure 4B** and **Supplementary Figure S3**). These results suggest that the AAGbox was acquired at a very early stage of embryophyte evolution, and is likely conserved in gymnosperms and angiosperms. On the other hand, monilophytes may have acquired another type of psbD promoter with a distinct cis element upstream of the −35 element.

### The AAG-Box Containing psbD Promoter Is not Associated with Light-Induced psbD Expression

In order to address whether the psbD LRP is responsible for light-induced transcription, we examined the light-induced expression of psbD transcripts in some embryophytes, including monilophytes, gymnosperms, and angiosperms. As shown in **Figure 5A**, psbD expression is clearly induced by light in A. thaliana (angiosperm). However, unexpectedly, no lightinduced psbD expression was detected in L. nobilis (basal angiosperm), P. thunbergii (gymnosperm), or Cycas revoluta (primitive gymnosperm), although they all have a well-conserved AAG-box containing psbD promoter. Similarly, psbD expression


is also not regulated by light in A. capillus-veneris (monilophyte), which does not have the conserved AAG-box in the psbD promoter. We further examined the light-mediated regulation of the AAG-box containing psbD promoter transcripts by primer extension analysis. As shown in **Figure 5B**, the abundance of the transcripts from the AAG-box containing psbD promoter was not regulated by light in L. nobilis or P. thunbergii (**Figure 5B**). These results suggest that the AAG-box containing psbD promoter is not directly associated with light-induced psbD expression.

Moreover, we examined light- and osmotic stress-induced psbD expression in a wide range of plants. In angiosperms, light-induced psbD expression was detected in a number of eudicots (C. sativus, A. thaliana, and L. sativa) and monocots (wheat and maize), whereas psbD expression is not regulated by light in basal angiosperms except for C. glaber (**Supplementary Figures S4–S6**). It is to be noted that psbD expression is activated by long-term illumination (12 h) in C. glaber (**Supplementary Figure S6**). In contrast, the osmotic stress-induced expression of psbD was detected only in eudicots. Expression of psbD was not activated by osmotic stress in monocots and basal angiosperms except for C. glaber (**Supplementary Figure S5**, **S6**). Furthermore, neither light nor osmotic stress-induced psbD expression was detected in gymnosperms, Gingko and Cycas (**Supplementary Figure S6**). Although psbD expression was not activated by moderate light (180 µmolm−<sup>2</sup> s −1 ) in Pinus (**Figure 5**), high light exposure (245 µmolm−<sup>2</sup> s −1 ) induced transient expression of psbD (**Supplementary Figure S6**). It is of note that gymnosperms and basal angiosperms have a wellconserved AAG-box containing psbD promoter. On the other hand, psbD expression is induced by light and/or osmotic stress in some monilophytes that lack the typical AAG-box containing psbD promoter. In addition, light and salt stress barely affect psbA expression in any of the plants examined (**Supplementary Figures S5–S8**). These results indicate that light and salt stressinduced transcription has evolved independently and multiple times during land plant evolution. Furthermore, the AAG-box in the psbD promoter is unlikely to be directly associated with the occurrence of light and salt stress-induced psbD expression.

### Gnetaceae Plants in Gymnosperms Have Lost the AAG-Box Containing psbD Promoter

Gnetaceae plants are a unique group of gymnosperms that have evolved a morphological system related to that of the angiosperms. The most upstream transcription initiation sites have been mapped at 317 and 114 bp upstream of the psbD translation start site of Gnetum gnemon and Ephedra sinica, respectively (**Figures 6A,B**). The psbD transcripts of gnetaceae have shorter 5<sup>0</sup> -UTR compared with the standard psbD transcripts in other plants. The upstream sequences of the psbD transcripts are well-conserved among gnetaceae. No AAGbox-like sequences are found upstream of the psbD transcripts in gnetaceae, whereas the gnetaceae psbD promoters possess −35- and −10-like sequences (**Figure 6C**). The similar sequences

are also found upstream of the psbD-C operon of Welwitschia mirabilis. Moreover, AAG-box-like sequences have not been identified in the trnT-psbD intergenic regions of gnetaceae. We suggest that the AAG-box containing psbD promoter was lost in gnetaceae during their evolution. It would be very interesting to determine whether psbD expression is dependent on light in gnetaceaes.

### DISCUSSION

The psbD and psbC genes are organized in a psbD-psbC operon, which is well-conserved among plants and cyanobacteria. Transcription from the psbD-psbC operon is mediated solely by PEP in green tissues. Multiple transcriptional start sites (TSS) generating mRNAs with heterogeneous 5<sup>0</sup> transcript leaders have been mapped on the psbD-psbC operon in several angiosperm plants, including tobacco (Yao et al., 1989), barley (Christopher et al., 1992), wheat (Wada et al., 1994), rice (To et al., 1996), and Arabidopsis (Hoffer and Christopher, 1997). Light activates the expression of some psbD-psbC mRNAs, whereas the accumulation of other mRNAs is not regulated by light. Analysis of the promoter sequences immediately upstream of the light-induced TSS identified an unusual PEP promoter consisting of a core promoter with a weakly conserved −35 element along with an upstream cis element termed the AAG-box (**Figure 1A**). This promoter is specifically activated by highirradiance blue light and UV-A (Christopher and Mullet, 1994), and is designated as the psbD light-responsive promoter (psbD LRP) or psbD blue light-responsive promoter (psbD BLRP). As shown in **Figure 1B**, the psbD LRP is also activated by various environmental stresses (Nagashima et al., 2004). On the other hand, other light-insensitive promoters mapped on the psbD-psbC operon are standard PEP promoters composed of well-conserved −10 and −35 elements.

In order to investigate psbD LRP evolution, we mapped promoters responsible for the expression of the psbD-psbC operon by primer extension analysis in 11 embryophytes at different evolutionary stages from liverworts to angiosperms. We considered the psbD transcripts with the most upstream terminus as the primary transcripts, except for P. thunbergii and E. hyemale, which have another promoter upstream of the potential AAG-box containing psbD promoter. All psbD promoters identified at the most upstream mRNA terminus

have the well-conserved −10 element (TATTCT) that is similar to the standard −10 element (TATAAT). On the other hand, the potential −35 elements of the psbD promoters are less conserved, suggesting the limited role of the −35 element in psbD promoter activity. In vitro transcription experiments have demonstrated that the −10 element is important for transcription from the AAG-box containing psbD promoter, but the −35 element is not essential for transcription in barley, rice, or wheat (To et al., 1996; Nakahira et al., 1998; Kim et al., 1999). It is considered that the −10 element is important for psbD promoter activity, but not the poorly conserved −35 element.

We found that the 19 bp AAG-box sequences of the psbD promoters are highly conserved among gymnosperms and angiosperms, including the basal angiosperm L. nobilis and primitive gymnosperm G. biloba. The consensus AAGbox sequence is "RAAGTAAGTRRACCTRACY," which is at least 80% identical in gymnosperms and angiosperms. The AAG-box is composed of two repeat sequences: AAGT and GACC/T repeats. Extensive analyses of the AAG-box containing psbD promoter structure using in vitro transcription systems have revealed that both the AAGT and GACC/T repeats are important for AAG-box containing psbD promoter activity (Kim and Mullet, 1995; To et al., 1996; Nakahira et al., 1998). Moreover, the AAG-box is also partially conserved in basal land plants such as lycophytes, hornworts, and mosses. We identified the shorter conserved AAG-box-like sequences in H. lucidula (lycophyte) and A. formosae (hornwort). The AAG-box-like sequences in lycophyta and hornworts retain a partially conserved GACC/T repeat, but lack AAGT repeats. Deletion of the AAGT repeat resulted in only a partial reduction of in vitro transcription activity of the AAG-box containing psbD promoter in wheat, suggesting that the GACC/T repeat is sufficient to mediate AAG-box containing psbD promoter activity (Nakahira et al., 1998). Thus, the AAG-box-like sequences may act as a transcription activation element in the psbD promoters of basal land plants. Considering the highly conserved −10 element among bryophytes and angiosperms, it is likely that the last common ancestor of bryophytes and spermatophytes likely already possessed an AAG box-containing psbD LRP (**Supplementary Figure S9**). The AAG box may have developed to take over the function of the −35 element and support the highlevel transcription activity of the AAG-box containing psbD promoter.

On the other hand, the AAG-box like sequences have not been identified in the psbD promoter of M. polymorpha (liverwort). The liverwort psbD promoter is composed of the typical −35 element (TTGAAA) and the −10 element (TATTCT) with standard spacing, suggesting that psbD is transcribed from a standard PEP promoter in liverworts. It is to be noted that the psbD promoter has a well-conserved −10 element, but lacks the AAG-box and standard −35 element in Chlamydomonas (Klein et al., 1992; Klinkert et al., 2005).

Monilophytes (ferns) have another type of psbD promoter that lacks the conserved AAG-box. Instead, 11-bp sequences upstream of the potential −35 element are well-conserved among standard (A. capillus-veneris) and primitive (P. nudum) monilophytes. However, it remains elusive whether the conserved sequences upstream of the psbD transcription stat site in monilophytes is required for psbD transcription. Interestingly, gnetaceae plants in gymnosperms have lost the AAG-boxcontaining psbD LRP, suggesting that the AAG-box containing psbD promoter is not essential for plant development.

It has also been shown that an AAG-box-binding factor (AGF) specifically binds to the AAGT repeat, that it is partially associated with the GACC/T repeats (Kim and Mullet, 1995; Kim et al., 1999), and that it activates transcription from the AAG-box containing psbD promoter (Wada et al., 1994; Kim and Mullet, 1995; Allison and Maliga, 1995; To et al., 1996; Nakahira et al., 1998; Kim et al., 1999). PTF1 (plastid transcription factor 1) is a basic helix-loop-helix DNA-binding protein, which binds to the AAG box and is involved in transcription from the psbD LRP in Arabidopsis (Baba et al., 2001). However, close orthologs of the Arabidopsis PTF1 have only been found in angiosperms, suggesting that PTF1 is responsible for AAG-box-dependent transcription at the AAG-box containing psbD promoter in angiosperms. However, the role of AGF in the light-dependent transcription remains to be elucidated.

In addition, reverse genetic analysis of sigma factors revealed that the AAG-box containing psbD promoter is specifically recognized by SIG5 in Arabidopsis (Nagashima et al., 2004; Tsunoyama et al., 2004). Transcription from the AAG-box containing psbD promoter is likely to be mediated by SIG5 containing PEP and activated by AGF that binds to the AAG-box. However, overexpression of SIG5 cannot activate transcription from the AAG-box containing psbD promoter in the dark (**Figure 1C**). Furthermore, photoreceptors including CRY1, CRY2, and PhyA are involved in the light-induced expression of the psbD-psbC operon in Arabidopsis (Thum et al., 2001a). Taken together, photoreceptor-mediated signaling may modify SIG5 or SIG5 import into the chloroplasts in a light-dependent manner and activate transcription at the AAG-box containing psbD promoter.

SIG5 orthologs have been identified in a number of angiosperms. Moreover, SIG5 has been identified as a gene that was abundant when water availability was low in gymnosperm Pseudotsuga menziesii (Hess et al., 2016). Furthermore, SIG5 orthologs have been identified in P. patens (Ichikawa et al., 2008) and Selaginella moellendorffii (XP\_002970534). It has been shown that PpSIG5 is involved in high-intensity light and circadian control of psbD expression in the moss P. patens (Ichikawa et al., 2004, 2008). Taken together, these results suggest that SIG5 plays a role in transcription from the psbD promoter consisting of the AAG-box or AAG-like element in embryophytes including mosses. On the other hand, an SIG5 ortholog was also identified in the liverwort M. polymorpha (Kanazawa et al., 2013). MpSIG5 is not necessary for light-dependent psbD expression in M. polymorpha (Kanazawa et al., 2013). MpSIG5 may have another role in chloroplast transcription in M. polymorpha that lacks the AAG-box containing psbD promoter. Furthermore, SIG5 orthologs have been identified in some charophytes such as K. flaccidum, but not in the green alga C. reinhardtii or in the primitive red alga C. merolae. It is likely that SIG5 was acquired in charophytes before the occurrence of AAG-box containing psbD promoter in basal embryophytes.

Transcripts from the AAG-box containing psbD promoter are characterized by a markedly long 5<sup>0</sup> -UTR. All AAG-box containing psbD promoter transcripts except for P. patens and M. polymorpha have a 5<sup>0</sup> -UTR longer than 500 nucleotides from the translation start site. Detailed mapping of the psbD transcripts in angiosperm plants revealed that no intron is present in the 5 0 -UTR and the AAG-box containing psbD promoter transcripts actually have a long 5<sup>0</sup> -UTR (Wada et al., 1994; Hoffer and Christopher, 1997; Kim et al., 1999). The 5<sup>0</sup> -UTR of chloroplast transcripts may be involved in the stability of the transcripts and/or efficiency of translation (Shiina et al., 1998). However, no conserved sequences have been identified in the 5<sup>0</sup> -UTR of embryophytes. Further characterization of the 5<sup>0</sup> -UTR would shed light on the role of the unusually long 5<sup>0</sup> -UTR of the AAG-box containing psbD promoter in most embryophytes,

This study suggest that the AAG-box containing psbD promoter appeared in basal embryophytes more than 450 million years ago, and the common ancestor of bryophytes and spermatophytes likely possessed an AAG box-containing psbD promoter. On the other hand, the −35 and −10 elements of the psbA and rbcL promoters are almost identical among liverworts and angiosperms (**Supplementary Figure S10**). It is suggested that ecological and/or physiological demands may have accelerated the evolution of the AAG-box containing psbD promoter in embryophytes. One of the unique characteristics of the AAG-box containing psbD promoter is light- and stressinduced transcription. However, extensive expression analysis of psbD in a variety of plants revealed that the light and/or stressinduced expression of the psbD gene developed independently in several plants. Thus, it is unlikely that that light and/or stress responses of the AAG-box containing psbD promoter are directly associated with AAG-box containing psbD promoter evolution. Recent studies demonstrated that ABA and the circadian rhythm regulate chloroplast AAG-box containing psbD promoter activity via the activation of SIG5 (Noordally et al., 2013; Yamburenko et al., 2015; Belbin et al., 2017) Further analysis of the role of the AAG-box containing psbD promoter in chloroplasts may shed light on AAG-box containing psbD promoter evolution.

### AUTHOR CONTRIBUTIONS

TS, SS, SM, and YN designed research. SS, MN, ShK, SaK, and YI performed research. TS, SS, and MN analyzed data. TS, SS, and MN wrote the paper. SS and MN have contributed equally to this work.

### FUNDING

This work was supported by JSPS and MEXT Grants- in-Aid for Scientific Research to TS (25291065, 15K14553, 15H01236, 17H05728, and 17H03968) and YN (15K14912), a grant from the Mitsubishi Foundation to TS, and a grant for high-priority study from Kyoto Prefectural University.

### ACKNOWLEDGMENTS

fpls-08-01186 July 11, 2017 Time: 15:44 # 11

We acknowledge Kyoto Botanical Garden for co-operation for plant sample collection. We also thank Dr. H. Tobe and Dr. M. H. Sato for helpful discussion.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2017.01186/ full#supplementary-material

FIGURE S1 | Mapping of the 5<sup>0</sup> ends of the psbD transcripts psbD transcripts were isolated from liverwort M. polymorpha (A), lycophyte H. lucidula (B), and monilophyte P. nudum (C). Primers used are indicated by numbers on the top of each lane. The size of the extension product is shown on the right. The position of primers and the size of the extension products are shown on the gene map. The deduced sites of the 5<sup>0</sup> -end of each transcript are shown as numbers in parentheses. Lane Xs show the results of unrelated samples. Lane C shows an experiment with the control RNA and primer provided by the manufacturer that produces an 87-base primer extension product.

FIGURE S2 | Mapping of the 5' ends of the psbD transcripts psbD transcripts were isolated from monilophyte E. hyemale (A), gymnosperm P. thunbergii (B), and angiosperm L. nobilis (C). Primers used are indicated by numbers on the top of each lane. The size of the extension product is shown on the left in (A) and on the right in (B,C). The position of primers and the size of the extension products are shown on the gene map. The deduced sites of the 5<sup>0</sup> -end of each transcript are shown as numbers in parentheses. Lane C shows an experiment with the control RNA and primer provided by the manufacturer that produces an 87-base primer extension product.

FIGURE S3 | A comparison of the potential psbD promoter sequences in monilophytes DNA sequences upstream of the psbD transcription initiation site of A. capillus-veneris and corresponding sequences of three other leptosporangiate ferns are compared with the psbD promoter sequences of two primitive ferns

### REFERENCES


(P. nudum and E. hyemale). Transcription initiation sites identified in this study are indicated by asterisks. Nucleotides that are identical to the Adiantum sequences are shown in red. The monilophytes-type upstream conserved sequences are indicated by a green box. The −35 and −10 elements are indicated by orange and yellow boxes, respectively.

FIGURE S4 | Northern blot analysis of psbD and psbA transcripts in eudicots. Plants were dark-adapted for 72 h (D) and illuminated for up to12 h (275 µmolm−<sup>2</sup> s −1 ; L). Osmotic stress was applied by 250 mM mannitol for 6 h. Light- or salt stress-induced psbD transcripts are indicated by red arrow heads. rRNA stained with EtBr was used as an RNA-loading control for the total RNA sample.

FIGURE S5 | Northern blot analysis of psbD and psbA transcripts in monocots. Plants were dark-adapted for 72 h (D) and illuminated for up to12 h (275 µmolm−<sup>2</sup> s −1 ; L). Osmotic stress was applied by 250 mM mannitol for 6 h. Light- or salt stress-induced psbD transcripts are indicated by red arrow heads. rRNA stained with EtBr was used as an RNA-loading control for the total RNA sample.

FIGURE S6 | Northern blot analysis of psbD and psbA transcripts in basal angiosperms. Plants were dark-adapted for 72 h (D) and illuminated for 12 h (275 µmolm−<sup>2</sup> s −1 ; L). Osmotic stress was applied by 250 mM mannitol for 6 h. Light- or salt stress-induced psbD transcripts are indicated by red arrow heads. rRNA stained with EtBr was used as an RNA-loading control for the total RNA sample.

FIGURE S7 | Northern blot analysis of psbD and psbA transcripts in gymnosperms. Plants were dark-adapted for 72 h (D) and illuminated for up to 12 h (275 µmolm−<sup>2</sup> s −1 ; L). Osmotic stress was applied by 250 mM mannitol for 6 h. Light- or salt stress-induced psbD transcripts are indicated by red arrow heads. rRNA stained with EtBr was used as an RNA-loading control for the total RNA sample.

FIGURE S8 | Northern blot analysis of psbD and psbA transcripts in monilophytes. Plants were dark-adapted for 72 h (D) and illuminated for up to 12 h (275 µmolm−<sup>2</sup> s −1 ; L). Osmotic stress was applied by 250 mM mannitol for 6 h. Light- or salt stress-induced psbD transcripts are indicated by red arrow heads. rRNA stained with EtBr was used as an RNA-loading control for the total RNA sample.

FIGURE S9 | Phylogenetic tree of AAG box. Alignments were undertaken with 19 bps sequences (from −54 to −36 of the transcription initiation site) of 18 plants from C. reinhardtii to A. thaliana.

FIGURE S10 | Comparion of psbA and rbcL promoters of terrestrial plants.

chloroplast transcription. Curr. Genet. 49, 333–340. doi: 10.1007/s00294-006- 0060-7


distinct groups of genes in tobacco plastids. EMBO J. 16, 4041–4048. doi: 10.1093/emboj/16.13.4041



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Shimmura, Nozoe, Kitora, Kin, Matsutani, Ishizaki, Nakahira and Shiina. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Arabidopsis thaliana mTERF10 and mTERF11, but Not mTERF12, Are Involved in the Response to Salt Stress

### Duorong Xu, Dario Leister and Tatjana Kleine\*

Plant Molecular Biology, Department Biology I, Ludwig-Maximilians-Universität München, Planegg-Martinsried, Germany

Plastid gene expression (PGE) is crucial for plant development and acclimation to various environmental stress conditions. Members of the "mitochondrial transcription termination factor" (mTERF) family, which are present in both metazoans and plants, are involved in organellar gene expression. Arabidopsis thaliana contains 35 mTERF proteins, of which mTERF10, mTERF11, and mTERF12 were previously assigned to the "chloroplast-associated" group. Here, we show that all three are localized to chloroplast nucleoids, which are associated with PGE. Knock-down of MTERF10, MTERF11, or MTERF12 has no overt phenotypic effect under normal growth conditions. However, in silico analysis of MTERF10, -11, and -12 expression levels points to a possible involvement of mTERF10 and mTERF11 in responses to abiotic stress. Exposing mutant lines for 7 days to moderate heat (30◦C) or light stress (400 µmol photons m−<sup>2</sup> s −1 ) fails to induce a phenotype in mterf mutant lines. However, growth on MS medium supplemented with NaCl reveals that overexpression of MTERF11 results in higher salt tolerance. Conversely, mterf10 mutants are hypersensitive to salt stress, while plants that modestly overexpress MTERF10 are markedly less susceptible. Furthermore, MTERF10 overexpression leads to enhanced germination and growth on MS medium supplemented with ABA. These findings point to an involvement of mTERF10 in salt tolerance, possibly through an ABA-mediated mechanism. Thus, characterization of an increasing number of plant mTERF proteins reveals their roles in the response, tolerance and acclimation to different abiotic stresses.

#### Keywords: Arabidopsis, chloroplast, nucleoid, mTERF, acclimation, stress, salt

### INTRODUCTION

Chloroplasts are of cyanobacterial origin (Raven and Allen, 2003) and harbor nowadays a reduced genome that mainly encodes proteins involved in photosynthesis and plastid gene expression (PGE). PGE is crucial for plant development and photosynthesis, but its regulation is only partially understood. This is largely because, although plastids still display characteristics of a prokaryoticlike structure of their genome, their gene expression machinery is much more elaborated compared to that of their cyanobacterial ancestor (reviewed in: Liere et al., 2011). Therefore, PGE requires plenty of proteins encoded in the nucleus that support transcription, splicing, trimming and editing of organellar RNAs, and regulate their translation (Schmitz-Linneweber and Small, 2008; Stern et al., 2010; Hammani et al., 2014; Tiller and Bock, 2014; Börner et al., 2015).

Edited by:

Julian Eaton-Rye, University of Otago, New Zealand

#### Reviewed by:

Victor Quesada, Universidad Miguel Hernández de Elche, Spain Ren Maozhi, Chongqing University, China

> \*Correspondence: Tatjana Kleine tatjana.kleine@lmu.de

#### Specialty section:

This article was submitted to Plant Cell Biology, a section of the journal Frontiers in Plant Science

Received: 20 March 2017 Accepted: 27 June 2017 Published: 14 July 2017

#### Citation:

Xu D, Leister D and Kleine T (2017) Arabidopsis thaliana mTERF10 and mTERF11, but Not mTERF12, Are Involved in the Response to Salt Stress. Front. Plant Sci. 8:1213. doi: 10.3389/fpls.2017.01213

Also the nucleus-encoded proteins of the mitochondrial transcription termination factor (mTERF) family regulate mitochondrial and PGE at diverse levels (Kleine and Leister, 2015). The mTERF proteins have been identified in both plants and metazoans (Linder et al., 2005). Human mTERF1, which is the first characterized mTERF, is one of four mammalian mTERF proteins, and was identified nearly 30 years ago as a factor that acts on transcription termination in mitochondrial extracts (Kruse et al., 1989). Its presumptive function as a transcription terminator (of heavy-strand transcripts) gave the family its name. More recently however, models have been suggested in which mTERF1 acts chiefly as a terminator of antisense transcription (Terzioglu et al., 2013) or in polar replication fork arrest (Shi et al., 2016). The true molecular function of mouse mTERF2 also remains unclear, with some reports suggesting that it binds to the same mitochondrial DNA region as mTERF1 and mTERF3 (Wenz et al., 2009), while another contends that the DNAbinding activity of mTERF2 is not sequence-specific (Pellegrini et al., 2009). Knock-out of Mterf3 in mice leads to embryonic lethality (Park et al., 2007), and conditional knockout of Mterf3 in the heart has identified a novel role for its protein product in the biogenesis of metazoan mitochondrial ribosomes (Wredenberg et al., 2013). Mterf4 knock-out mice are also embryonic lethal (Camara et al., 2011). Interestingly, human mTERF4 forms a complex with NSUN4, which is required for assembly of the small and large ribosomal subunits of the mitochondrial ribosome (Metodiev et al., 2014). Consequently, while the function for mTERF2 remains to be clarified, the remainder of the mammalian mTERFs do not support transcription termination, as it is suggested by their notation, but seem to take part in antisense transcription termination and ribosome biogenesis.

The number of mTERF family members has increased to approximately 30 throughout the evolution of land plants (Kleine, 2012), but information on their functions is only beginning to emerge. Most of the 35 A. thaliana mTERF proteins (mTERF1-mTERF35; Kleine, 2012) are localized to chloroplasts and/or mitochondria (Babiychuk et al., 2011), and seven of them (mTERF1, -4, -5, -6, -9, -15, and -18) have been functionally investigated in more detail (reviewed in: Kleine and Leister, 2015; Quesada, 2016). Essential functions of mTERF proteins in plant development are revealed by the effects of complete inactivation of three MTERF genes: A. thaliana mutants devoid of SOLDAT10 (SINGLET OXYGEN-LINKED DEATH ACTIVATOR10)/mTERF1 (Meskauskiene et al., 2009) or BSM (BELAYA SMERT)/RUG2 (RUGOSA2)/mTERF4 (Babiychuk et al., 2011; Quesada et al., 2011) are arrested in embryo development, and knock-out mterf6-2 plants are albinotic and stop growing after 2 weeks (Romani et al., 2015). Moreover, the dissection of mterf mutants supports an involvement of plant mTERFs in responses to abiotic stress (reviewed in: Kleine and Leister, 2015; Quesada, 2016). Indeed, SOLDAT10 (Meskauskiene et al., 2009) and SUPPRESSOR OF hot1– 4 1 (SHOT1; Kim et al., 2012) were isolated in forward genetic screens for loci that influence responses to abiotic stress. The hot1-4 mutant is a dominant-negative allele of the heat-shock protein gene HSP101. SHOT1/mTERF18 is a mitochondrial protein and the shot1-1 missense mutant and the shot1-2 T-DNA insertion mutant each suppress the heat hypersensitivity of hot1-4 plants. Moreover, other heat-sensitive mutant phenotypes are also suppressed by shot1-2, and shot1- 2 single mutants display a higher heat tolerance (Kim et al., 2012). SOLDAT10 is localized to chloroplasts, and plants homozygous for a weaker soldat10 allele suffer from mild photooxidative stress already in low-light conditions; this results in turn in a stress acclimation response, which appears to confer improved hardiness against a combination of high-light and low-temperature stress (Meskauskiene et al., 2009). Other mterf mutants are also linked to stress responses. For example, mda1 (mterf5), and mterf9 seedlings are less susceptible to osmotic and salt treatments, which might be linked to their decreased sensibility to ABA (Robles et al., 2012, 2015). Furthermore, the rug2-1 mutant is abnormally sensitive to temperature stress. At 26◦C, rug2-1 homozygotes undergo growth arrest, whereas at 16◦C this growth phenotype is not expressed (Quesada et al., 2011).

A co-expression network for all MTERF genes (26 out of 35) which were present on the Affymetrix ATH1 genome array has been constructed (Kleine, 2012). The resulting clusters and information related to the subcellular locations of the proteins that are encoded by genes co-expressed with each MTERF gene were then used to assign the mTERFs into five groups, referred to as the "chloroplast," "chloroplast-associated," "mitochondrial," "mitochondrion-associated," and the "low expression" clusters.

In the present study, we characterized the members of the "chloroplast-associated" group, which comprises mTERF10 (AT2G34620), mTERF11 (AT3G18870), and mTERF12 (AT4G09620). The sub-chloroplast localization of mTERF10, -11, and -12 was defined by fluorescence microscopy of mTERF-GFP fusions and an RFP fusion protein (as a control for nucleoid localization). Lines with altered MTERF10, MTERF11, and MTERF12 levels did not display phenotypes under normal growth conditions. In silico analyses with the eFP browser and Genevestigator were conducted, which pointed to an involvement of these three mTERFs in abiotic stress responses. To follow this up, the mutant lines were exposed to moderate heat (30◦C), high light (400 µmol photons m−<sup>2</sup> s −1 ), or salt (175 mM NaCl) stress, and subjected to ABA treatment. The mterf10, -11, and -12 mutant lines responded to heat and high light stress like the wild type (WT). However, lack of mTERF10 or mTERF11 led to enhanced or reduced sensitivity to salt, respectively, while overexpression of MTERF10 rendered seedlings more tolerant than WT to both salt and ABA.

### MATERIALS AND METHODS

### Plant Material and Growth Conditions

The mutants mterf10-1 (SAIL\_12A03), mterf10-2 (SALK\_097699), mterf11-1 (FLAG\_357F09), mterf11-2 (GABI\_211D05), and mterf12-1 (GABI\_407E04) were identified in the SIGnAL database (Alonso et al., 2003), the abi4-1 mutant was ordered from The European Arabidopsis Stock Centre (NASC; ID N8104). All mutants are in the Col-0 background except of mterf11-1 which is a WS line.

Arabidopsis thaliana plants were grown in long-day conditions (16 h light/8 h dark) on potting soil (Stender). Plants were illuminated with HQI Powerstar 400 W/D lamps and a fluence rate of approximately 100 µmol photons m−<sup>2</sup> s −1 . To accomplish salt and ABA stress experiments, seedlings were grown on plant agar (Sigma-Aldrich) containing half-strength MS medium and 1.5% (w/v) sucrose at 22◦C under 100 µmol photons m−<sup>2</sup> s <sup>−</sup><sup>1</sup> provided by white fluorescent lamps under continuous light or long-day conditions. For salt stress experiments, MS medium was supplemented with 125 mM or 175 mM NaCl as indicated. For ABA experiments, MS medium was supplemented with 1 µM ABA.

### Nucleic Acid Extraction

For DNA isolation, leaf tissue was homogenized in extraction buffer containing 200 mM Tris/HCl (pH 7.5), 25 mM NaCl, 25 mM EDTA, and 0.5% (w/v) SDS. After centrifugation, DNA was precipitated by adding isopropyl alcohol. After washing with 70% (v/v) ethanol, the DNA was dissolved in distilled water.

For RNA isolation, frozen tissue was ground in liquid nitrogen. Following the addition of TRIZOL (Invitrogen) and chloroform according to the manufacturer's instructions, RNA was precipitated from the aqueous phase with isopropyl alcohol, then washed with 70% (v/v) ethanol, and dissolved in RNasefree water. Concentration and purity of RNA samples were determined spectroscopically in a GeneQuant pro RNA/DNA Calculator (GE Healthcare Europe GmbH). Isolated RNA was stored at −80◦C until further use.

### cDNA Synthesis and Real-Time PCR Analysis

cDNA synthesis and real-time PCR analysis were performed as outlined before (Voigt et al., 2010). All reactions were done in triplicate on three biological replicates. The target genes and the respective primers, are listed in **Supplementary Table S1**. The RCE1 gene was used as an internal reference in other studies (Voigt et al., 2010; Romani et al., 2015). RCE1 transcript levels are not changed upon diverse conditions, especially not under diverse stress conditions including lincomycin and norflurazon treatment which affect organellar gene expression.

### RNAi, Overexpression and Intracellular Protein Localization

To reduce MTERF12 mRNA levels by RNAi, a 145 bp fragment was amplified from genomic DNA with the primer pair AT4G09620-GST-attB1 and –attB2 (see **Supplementary Table S1**). The gel-purified PCR product was used for BP and LR Clonase reactions (GATEWAY Cloning; Invitrogen) which led to the final construct pB7GWIWG2/MTERF12 (for pB7GWIWG2, see Karimi et al., 2002). For overexpression and localization studies of mTERF10, mTERF11 and mTERF12, cDNAs encompassing the coding regions were amplified by PCR (see **Supplementary Table S1** for primer information). Notably, in our Col-0 strain, MTERF11 has an additional triplet (CAT; coding for histidine) inserted after nucleotide 27 (relative to the start codon) compared with the coding sequence from The Arabidopsis Information Resource (TAIR; www. arabidopsis.org). MTERF10, MTERF11, and MTERF12 were cloned by GATEWAY technology (see above) into pB7FWG2 to generate fusions with enhanced GFP (eGFP), expression of which is controlled by the Cauliflower mosaic virus 35S promoter. For RAP-RFP fusions, the pENTR/RAP plasmid (Prof. Jörg Nickelsen, LMU Munich) was introduced into p2GWR7 by GATEWAY cloning. For overexpression of mTERF10, MTERF10 was introduced by classical cloning with the NcoI restriction enzyme into pCAMBIA1302. For RNAi experiments with MTERF12 and overexpression of mTERF10 and mTERF11, the plasmids pB7GWIWG2/MTERF12, pCAMBIA1302/MTERF10, and pB7FWG2/MTERF11 were independently transferred into Agrobacterium tumefaciens, and A. thaliana (ecotype Col-0 for MTERF10 overexpression and MTERF12 RNAi; ecotype WS for MTERF11 overexpression) plants were transformed by the floral-dip method (Clough and Bent, 1998). After seed set, transgenic plants were selected on the basis of their resistance to BASTA (pB7GWIWG2/MTERF12 and pB7FWG2/MTERF11) or hygromycin (pCAMBIA1302/MTERF10), respectively.

For fluorescence visualization, leaves of 3-week-old Col-0 plants grown on MS medium were cut into small pieces and incubated for 16 h at 24◦C in the dark in a protoplasting solution (10 mM MES, 20 mM CaCl2, 0.5 M mannitol (pH 5.8), 0.1 g ml−<sup>1</sup> macerozyme (Duchefa), 0.1 g ml−<sup>1</sup> cellulase (Duchefa). After isolation and transformation of protoplasts as described (Dovzhenko et al., 2003), preparations were examined with a Fluorescence Axio Imager microscope (Zeiss). Fluorescence was excited with the X-Cite Series 120 fluorescence lamp (EXFO) and images were collected at 500–550 nm (eGFP fluorescence), 570– 640 nm (RFP fluorescence) and 670–750 nm (chlorophyll autofluorescence).

### Chlorophyll a Fluorescence Measurements

In vivo chlorophyll a fluorescence of whole plants was recorded using an imaging chlorophyll fluorometer (ImagingPAM, Walz GmbH, Effeltrich, Germany). Plants were dark adapted for 15 min and then exposed to a pulsed, blue measuring light (1 Hz, intensity 4) and a saturating light flash (intensity 5) to determine the maximum fluorescence F<sup>m</sup> and the ratio (Fm-F0)/F<sup>m</sup> = Fv/Fm.

### Computational Analyses

Protein sequences were retrieved from the National Center for Biotechnology Information (NCBI; http://www.ncbi.nlm.nih. gov/) and The Arabidopsis Information Resource (TAIR; http:// www.arabidopsis.org). Amino acid sequences were aligned using the ClustalW program (http://www.ebi.ac.uk/clustalw; Chenna et al., 2003). The unrooted tree was constructed with the Phylip server Mobyle at the Pasteur Institute (http://mobyle.pasteur.fr/ cgi-bin/portal.py#welcome).

### RESULTS

### All Members of the Chloroplast-Associated mTERF Cluster Are Localized to Nucleoids

The localizations of almost all A. thaliana mTERF proteins have been investigated by fluorescence microscopy of mTERF-GFP fusions transiently expressed in isolated protoplasts, and in guard cells of transgenic plants (Babiychuk et al., 2011). These data indicated that mTERF10, -11, and -12 are targeted to chloroplasts. To confirm these results and if possible define the precise locations of the proteins within the chloroplast, the eGFP fluorescence of mTERF10-, mTERF11-, or mTERF12-eGFP fusions, transiently overexpressed in Col-0 protoplasts, was monitored. Localization of all three fusion proteins to chloroplasts was confirmed (**Figure 1A**). However, the fluorescence signals were not uniformly distributed, but

FIGURE 1 | Localization of mTERF10, mTERF11 and mTERF12. (A) Fluorescence microscopy of A. thaliana protoplasts transiently expressing mTERF10, mTERF11 or mTERF12 fused to eGFP (mTERF10-eGFP, mTERF11-eGFP and mTERF12-eGFP). (B) To visualize nucleoids, protoplasts were co-transformed with a RAP-RFP fusion in combination with mTERF10-eGFP, mTERF11-eGFP, or mTERF12-eGFP. The eGFP fluorescence is shown in green (GFP), RFP fluorescence in cyan (RFP), autofluorescence of chloroplasts in red (Auto). The scale bars correspond to 10 µm.

appeared as small dots in the chloroplasts. The size and distribution of these dots were suggestive of nucleoids, which are associated with PGE (Majeran et al., 2012). Localization of the A. thaliana RNA-binding protein RAP to nucleoids was previously established using a transiently expressed RAP-eGFP fusion (Kleinknecht et al., 2014). Therefore, Col-0 protoplasts were co-transformed with a RAP-RFP fusion in combination with mTERF10-eGFP, mTERF11-eGFP, or mTERF12-eGFP. Indeed, for each mTERF-eGFP construct, signals were found in dots together with the RFP signal. Merging of both signals confirmed colocalization of the mTERF10, -11, and -12 fusions with RAP, and therefore localization of all three mTERFs to nucleoids (**Figure 1B**). It is noteworthy, that especially mTERF12—and a minor fraction of RAP—tend to be localized in the chloroplast stroma when both mTERF12-eGFP and RAP-RFP are expressed together in protoplasts.

### Identification and Phenotypic Analysis of Mutants for the MTERF10, MTERF11, MTERF12 Loci

To obtain insight into the physiological functions of mTERFs 10, 11, and 12, T-DNA insertion lines were identified in the SIGnAL database (Alonso et al., 2003). The insertions were confirmed by PCR (**Figure 2A**) and homozygous lines were selected. In the mutants mterf10-1 (SAIL\_12A03) and mterf10- 2 (SALK\_097699) the T-DNAs are inserted in the 5′ UTR and the second exon, respectively (**Figure 2B**). The mterf11-1 (FLAG\_357F09) and mterf11-2 (GABI\_211D05) mutants both have their T-DNA insertion in the gene's single exon. For MTERF12, only one insertion line could be identified (mterf12- 1, GABI\_407E04), which contains a T-DNA in the promoter region (**Figure 2B**). To repress the MTERF12 gene by RNAi, Col-0 lines were generated that contained constructs with an inverted repeat of a fragment spanning the first exon and a part of exon 2 of MTERF12 (**Figure 2A**) which was under control of the constitutive Cauliflower mosaic virus 35S promoter.

**Figure 2C** shows the numbers and distributions of mTERF domains in the mTERF10, -11, and -12 proteins. In mTERF10 and mTERF11, six and five mTERF motifs are predicted by the Simple Modular Architecture Research Tool SMART (http://smart.embl-heidelberg.de/). One mTERF motif has been predicted for mTERF12 (our previous results, and see also Supporting Information of Babiychuk et al., 2011), but this domain is not annotated anymore with confidence by the SMART tool (http://smart.embl-heidelberg.de/smart/ show\_motifs.pl?ID=Q93ZZ2\_ARATH). Thus, the classification of mTERF12 as an mTERF protein must be regarded as uncertain.

All mutants are in the Col-0 background except of mterf11- 1 which is a WS line. Hence, in all following experiments, mterf11-1 was compared with WS, while Col-0 was used as the WT standard for the other lines. Real-time PCR analysis was employed to determine the extent of repression of MTERF transcripts in the different mutant lines (**Figure 3A**). In 3 week-old mterf10-1 and mterf10-2 plants, MTERF10 transcript levels were reduced to 29 and 4% of WT, respectively. To determine MTERF11 transcript levels, primer pair A was chosen to detect transcripts initiated 5′ of the T-DNA insertions (**Figure 2A**). Using this set-up, MTERF11 transcript levels were found to be unchanged (mterf11-1) and nearly 6-fold induced (mterf11-2) relative to their WT (**Figure 3A**). In the mterf11- 1 allele (which is FLAG\_357F09), the T-DNA of the pGKB5 vector integrated in the 5′LB–T-DNA–RB3′ direction. It is of note here that the pGKB5 vector used to generate the FLAGdb T-DNA insertion line collection contains the 35S promoter on the LB side (Samson et al., 2002). The 35S promoter drives the expression of PHOSPHINOTRICIN ACETYL TRANSFERASE (PAT) used to select transgenic plants, and the PAT transcripts are terminated by the G7 terminator. It was already shown with two independent FLAG lines as an example that the G7 terminator can be an inefficient terminator in the context of the pGKB5 vector, allowing transcription to continue through and beyond the terminator sequence (Ulker et al., 2008). However, real-time PCR carried out with a primer pair covering the region 3′ of the T-DNA insertion detected greatly reduced MTERF11 transcript levels in the mterf11 mutants: 0.09% of WT in mterf11-1 and 0.01% in mterf11-2 (**Figure 3A**). MTERF12 transcript levels were not affected in the mterf12- 1 mutant (**Figure 3A**). Therefore, MTERF12 RNAi lines were tested for their ability to repress MTERF12 gene expression. Six independent lines were screened, but the most effectively repressed lines, mterf12i-1 and mterf12i-2, still retained 32% and 59% of WT (Col-0) amounts of MTERF12 transcripts, respectively (**Figure 3A**). Under normal growth conditions, all identified mutant lines were phenotypically indistinguishable from WT (**Figure 3B**). To look for subtle photosynthetic phenotypes, the maximum quantum yield of photosystem II (Fv/Fm) was measured in Col-0, WS and all mterf mutants (**Figure 3B**), but no deviations in this parameter were detected in the mutants.

To summarize, the expression of all mTERF motifs should be strongly reduced in the mterf10 mutants (particularly mterf10- 2), while the mterf11 mutants produce truncated transcripts. Assuming the latter are translated, the protein products would lack the last two mTERF domains (mterf11-1) or mTERF domain 5 only (mterf11-2) (**Figure 2C**). In the mterf12i lines, transcripts including the single putative mTERF domain were—at best reduced to one-third of Col-0 levels. At all events, none of the mterf mutant lines display any obvious phenotype under normal growth conditions.

### Phylogenetic Position of the mTERF10, -11, and -12 Proteins

Because the mterf10, -11, and -12 mutant lines lacked a clear phenotype under normal growth conditions (**Figure 3**), we asked whether this might be attributable to functional redundancy within the mTERF family. Several MTERF genes have undergone tandem duplications (on chromosome 1) and one block duplication (AT4G19650 and AT5G45113; Kleine, 2012). But neither MTERF12 nor MTERF10 or MTERF11 originated from a duplication event, so we can exclude the possibility of

(Continued)

### FIGURE 2 | Continued

sequences around the T-DNA insertion. The combination of RP and T-DNA left border primer (LB) was used for the verification of the T-DNA insertion. (B) Schematic representation and T-DNA tagging of the MTERF10 (AT2G34620), MTERF11 (AT3G18870), and MTERF12 (AT4G09620) loci. Exons (black boxes), introns (black lines) and the 5′ and 3′ UTRs (gray boxes) are shown. Numbers are given relative to the transcription start site of the gene loci. Locations and orientation of T-DNA insertions are indicated, as deduced from RP + LB PCR products shown in (A) which were subsequently sequenced. Note that the insertions are not drawn to scale. Furthermore, the location of the MTERF12 RNAi-directed sequence is indicated as a dashed line. (C) Schematic representation of mTERF10, mMTERF11, and mTERF12 proteins. The numbers and locations of mTERF domains are shown as white boxes. The relative positions of T-DNA and RNAi tagging are indicated.

mutant plants grown under long-day (16/8 h) light conditions. The maximum quantum yield of PSII (Fv/Fm) was measured with an ImagingPAM fluorometer.

protein redundancy arising from gene duplication. To obtain an impression of the overall degree of sequence similarity within the mTERF protein family, we constructed a phylogenetic tree which included in addition to A. thaliana mTERFs, mTERF members from the green alga Chlamydomonas reinhardtii, Homo sapiens, Mus musculus, Drosophila melanogaster, and other organisms (**Figure 4**). The tree reveals four main clades. The mTERF members of C. reinhardtii, H. sapiens and M. musculus are all in the same clade. The majority of A. thaliana mTERFs form clade I which includes mTERF10 and -11, while mTERF12 along with five other A. thaliana mTERFs and one D. melanogaster mTERF

constitute clade II. The mTERF10, -11, and -12 proteins are most closely related to mTERF1/SOLDAT10 (Meskauskiene et al., 2009), mTERF4/BSM/RUG (Babiychuk et al., 2011; Quesada et al., 2011) and mTERF15 (Hsu et al., 2014) proteins. Mutants for each of these three display phenotypes under normal growth conditions and have been shown to be involved in PGE or mitochondrial gene expression. Moreover, levels of sequence identity/similarity between mTERF10 and mTERF1 (38/68% over a stretch of 240 amino acids), mTERF11 and mTERF4 (26/41% over a stretch of 222 amino acids) and mTERF12 and mTERF15 (29/53% over a stretch of 77 amino acids), respectively, are

### FIGURE 4 | Continued

of four main clades (I–IV). Clade I encompasses proteins encoded by a tandem gene cluster on A. thaliana chromosome 1 and several other A. thaliana mTERF proteins. In clade III, C. reinhardtii mTERFs are grouped together with animal mTERF proteins. Clade IV comprises mTERF proteins from diverse species including paramecium, sea urchin (P. livides), parasites (P. falciparum and B. bovis) together with mTERFs from Drosophila. The mTERF proteins mTERF10 and -11 (highlighted in large, bold letters) form clade I together with 25 other mTERF proteins, while mTERF12 (also highlighted in big, bold letters) is assigned to clade II, together with five other A. thaliana mTERFs and one Drosophila mTERF. C, chloroplast; M, mitochondrion; N, nucleus; Y, cytosol; O, other.

noteworthy for the mTERF10/mTERF1 pair, but negligible for the other two.

### Changes in MTERF Transcript Levels in Response to Abiotic Stresses

To gain deeper insights into the functions of mTERF10, -11, and -12, their mRNA expression patterns were analyzed. Coexpression analysis of 26 MTERF genes and their corresponding gene ontology annotations have already been reported (Kleine, 2012). However, that study was designed to provide a global classification. Hence subsequent Genevestigator analyses only dealt with the numbers of conditions/treatments that altered MTERF gene expression. In the present study, we extracted MTERF transcript levels from the Arabidopsis eFP browser (http://bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi) with "Abiotic Stress" as a data source (Winter et al., 2007). In these experiments, 18-day-old plants were subjected to different stresses, and samples were taken over a time course of 24 h from stress-treated and control plants. We calculated the relative changes in MTERF10, -11, and -12 transcript levels from plants exposed to drought, high salt, heat, or cold compared to control conditions (**Figure 5A**). Because mda1 (mterf5) and mterf9 mutants are known to exhibit altered stress responses (Robles et al., 2012, 2015), MTERF5 and -9 were included for reference. Under drought and heat stress, transcript levels of all investigated MTERF genes were only moderately changed (**Figure 5A**). With a 3-fold rise after 1 h of heat stress (MTERF10) and an approximately 0.3-fold change (MTERF5 and 9), those transcripts were the most responsive. Under salt and cold treatment, MTERF transcript levels tended to be reduced. Under both salt and cold stress, MTERF10 and MTERF11 levels were most responsive, and especially after 24 h of cold treatment MTERF5, 10, 11, and 12 transcript levels were reduced (**Figure 5A**). To confirm these data and to find other conditions under which the MTERFs of interest might be regulated, the Genevestigator Perturbations Tool (https:// genevestigator.com/gv; Hruz et al., 2008) was employed on all deposited A. thaliana ATH1 arrays together with a 2-fold change filter and a p-value of < 0.05. An overview of all changes in MTERF10, MTERF11, and MTERF12 mRNA levels in response to perturbations (relative to untreated controls) can be found in **Supplementary Figures S1**–**S3**. In **Figure 5B** selected conditions are shown which are associated with changes in temperature and light, and with salt and drought stress conditions. Levels of MTERF10 mRNA were most susceptible to perturbation, being induced by light, raised after germination, and strongly reduced under drought conditions and various cold and high-light regimes, and on exposure to salt or ABA (**Figure 5B**).

### Knockdown of MTERF10 or MTERF11 Alters Sensitivity to Salt Stress

To experimentally probe the involvement of mTERFs in stress responses, 3-week-old WT and mterf10, -11, and - 12 mutant plants grown under standard conditions (22◦C at 100 µmol photons m−<sup>2</sup> s −1 ) were exposed for 7 days to moderate temperature stress (30◦C, at a fluence rate of 100 µmol photons m−<sup>2</sup> s −1 ) or moderate light stress (400 µmol photons m−<sup>2</sup> s −1 at a temperature of 22◦C). After 7 days of moderate temperature stress, the leaf petioles of WT and all mterf mutants were shortened, but otherwise all plants looked healthy (**Figure 6A**). After 3 days, Fv/F<sup>m</sup> was slightly reduced in all mterf11 and mterf12 mutant lines, but was restored to normal levels after 7 days (**Figures 6A,B**). Also after 7 days of moderate light stress, the leaf petioles of WT and all mterf mutants were shortened—albeit to a lesser extent. Furthermore, the edges of older leaves in all lines began to show signs of necrosis (**Supplementary Figure S4A**). After 1 h of moderate light stress, Fv/F<sup>m</sup> was slightly reduced in all lines (**Supplementary Figure S4B**). This reduction continued in the mterf11-1 mutant after 2 and 4 h, but all lines recovered to the initial Fv/F<sup>m</sup> values after 96 h (**Supplementary Figure S4B**).

After 3 and 24 h of salt stress, MTERF5 and MTERF9 transcript levels were reduced to half of those in control conditions (**Figure 5A**), and indeed, mda1 (mterf5) and mterf9 seedlings are less sensitive to salt and osmotic stresses (Robles et al., 2012, 2015). Because MTERF10 and MTERF11 transcripts were reduced to an even larger extent than MTERF5 and MTERF9 RNAs following exposure to salt stress for 6 and 24 h (**Figure 5A**), we asked whether inactivation of MTERF10 or MTERF11 might enable the mutant plants to better tolerate salt stress. To this end, WT and mterf mutant lines were germinated on MS medium (control) and MS medium supplemented with 125 mM or 175 mM NaCl, and germination rates were scored after different time points (**Figure 7A**). All lines germinated to nearly 100% on the control MS medium. Germination rates of all lines with a Col-0 background grown for 48 h on MS medium supplemented with 125 mM NaCl or for 72 h on medium supplemented with 175 mM NaCl were very similar (**Figure 7A**). In the aforementioned conditions, germination rates of Col-0 seeds were approximately 60% (**Figure 7A**). The germination rates of mterf10-1 and -2, mterf11-2 and all mterf12 seeds were all lower (ranging from 26 to 47%) than those of Col-0 seeds. However, after 96 h on MS medium supplemented with 175 mM NaCl the germination rates of mterf12 seeds (76 to 87%) were comparable to that of Col-0 seeds (84%). Interestingly, mterf10-1, mterf10-2, and mterf11-2 still displayed enhanced sensitivity to salt inhibition, with germination rates of 54, 55, and 61%, respectively (**Figure 7A**). WS seeds were

FIGURE 5 | In silico analyses of changes in levels of MTERF transcripts in response to abiotic stresses. (A) MTERF transcript levels were extracted from the Arabidopsis eFP Browser (http://bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi) with "Abiotic Stress" as a data source (Winter et al., 2007). Plant material from stress-treated and control plants was analyzed over a time course of 24 h. Here, the expression values are reported relative to the corresponding transcript levels in control conditions. (B) The Genevestigator Perturbations Tool (https://genevestigator.com/gv; Hruz et al., 2008) was applied to all available A.thaliana microarrays in combination with the 2-fold change filter and a p-value of < 0.05. Shown here is a selection of conditions related to abiotic stresses. Conditions were ordered according to the magnitude of the relative change in MTERF10 mRNA (from high to low). An overview of all transcriptional responses of MTERF10, MTERF11, and MTERF12 to perturbations can be found in Supplementary Figures S1–S3.

very susceptible to salt stress and failed to germinate under the conditions used to investigate the Col-0 descendant lines. For this reason, a milder salt stress treatment was applied to all lines with a WS background. Still, after 72 h growth on MS medium supplemented with 125 mM, the germination rate of WS was only 7% and raised to 48% after 96 h (**Figure 7A**). Although the mterf11-2 mutant was more sensitive to salt stress compared to Col-0, the germination rates of mterf11-1 seeds were comparable to their corresponding WT (WS; **Figure 7A**).

To ascertain whether the altered activity of MTERF10 was indeed responsible for the salt-stress phenotypes and whether overexpression of MTERF11 might lead to enhanced saltstress tolerance, 35S:MTERF10:MGFP5 and 35S:MTERF11:EGFP constructs were introduced into Col-0 and WS, respectively, to generate oe-mTERF10 and oe-mTERF11 lines. Although MTERF10 mRNA levels were only approximately 2.3-fold higher in oe-mTERF10-1 and oe-mTERF10-2 lines than in Col-0 (**Figure 8A**), these lines—with germination rates of approximately 85 and 95%, respectively, after 48 h on MS medium with 125 mM NaCl and 72 h on MS medium with 175 mM NaCl—were nevertheless resistant to the deleterious effect of salt (**Figure 7A**). This confirms that the salt sensitivity of mterf10-1 and mterf10-2 mutants is indeed caused by knockdown of the MTERF10 gene. Moreover, we identified three oemTERF11 lines that displayed a high diversity of MTERF11 transcript overaccumulation which ranged from 12- to 117 fold (**Figure 8B**). Two of these lines were challenged with salt stress, and actually displayed much higher germination rates than WS and therefore, enhanced salt stress tolerance (**Figure 7A**).

#### FIGURE 7 | Continued

Seed germination was investigated on MS medium in the absence and presence of 125 and 175 mM NaCl. Radicle emergence was scored after indicated time points. Germination rates were calculated relative to the number of total seeds. (B) Phenotypes of 7-day-old WT and mutant seedlings were germinated on either MS medium supplemented with 125 mM NaCl (grown for 7 days on 125 mM NaCl), or on MS medium and transferred directly after radicle emergence for 12 days to MS medium supplemented with 125 mM NaCl (transferred for 12 days to 125 mM Nacl) or 175 mM NaCl (transferred for 12 days to 175 mM NaCl), respectively. The root lengths (C) and cotyledon greening rates of seedlings grown on 125 mM NaCl—displayed as the ratio of the number of green cotyledons to the total number of cotyledons (D) were determined after 7 and 5 days, respectively. The data in (A,C,D) represent mean values ± SD of three independent experiments, each performed with at least 100 seeds per treatment and genotype. Statistically significant differences (t-test; p < 0.05) between WT (Col-0 or WS) and corresponding mutant lines are indicated by an asterisk (black for Col-0; red for WS).

To investigate this further, the performance of mterf mutants and mTERF overexpression lines was investigated during postgermination development. As shown in **Figures 7B,C**, the root lengths of mterf10-1 and -2 seedlings challenged with 125 mM NaCl were significantly shorter compared to Col-0, while the root lengths of oe-mTERF10 seedlings were longer (**Figure 7C**). Moreover, compared to WS, overexpression of MTERF11 results in longer root lengths, reflecting findings of the germination rates (**Figure 7A**). In addition, cotyledon greening—displayed as the ratio of the number of green cotyledons to the total number of cotyledons—of mterf10-1 and -2 seedlings was delayed by salt stress, while in contrast, overexpression of MTERF10 or MTERF11 enabled seedlings to display higher cotyledon greening rates than their corresponding wild types (**Figure 7D**).

ABA operates as a signal during developmental processes including seed germination, and moreover, in response to abiotic stresses including salt stress (Christmann et al., 2006). Furthermore, A. thaliana mutants in which the ABI4 (ABSCISIC ACID INSENSITIVE4) gene has been inactivated are more salt tolerant than WT (Quesada et al., 2000; Shkolnik-Inbar et al., 2013). To investigate whether reduced MTERF transcript levels in the mterf10, -11, and -12 mutant lines or overexpression of mTERF10 or mTERF11 alter ABA sensitivity, wild-type, mterf, oe-mTERF10, oe-mTERF11 and—as a control—abi4-1 mutant seedlings were grown on MS supplemented with 1 µM ABA, and germination rates were scored after 96 h (lines with a Col-0 background) and 120 h (lines with a WS background). With a 69% germination rate, the control line abi4-1 germinated better than Col-0 (49%; **Figure 9A**). All mterf12 lines displayed a slightly, but not significantly, higher germination rate than Col-0, but mterf10 and mterf11 lines were as sensitive as Col-0 to ABA stress. Importantly, especially oe-mTERF10 lines were also less susceptible to ABA stress (**Figure 9A**), like they were to salt stress (**Figure 7A**). After 120 h on 1 µM ABA, WS germinated to 32%, and both mterf11-1 and oe-mTERF11 lines displayed even lower germination rates (**Figure 9A**).

To follow this up, the phenotypes of seedlings grown on 1 µM ABA were examined after 7 days. Col-0 and WS seedlings displayed short roots and cotyledons had only started to emerge (**Figure 9B**). In contrast, abi4-1 seedlings had longer roots and fully expanded cotyledons. All mutant lines with reduced MTERF10, -11, or -12 transcript levels showed the same behavior as the wild types. However, the cotyledon phenotype of the oemTERF10 lines was comparable to that of the abi4-1 mutant (**Figure 9B**). This was also manifested in the higher relative cotyledon greening rate of oe-mTERF10 lines (25 and 30%) compared to Col-0 (**Figure 9C**).

It appears that challenging Arabidopsis seedlings with ABA or salt stress under continuous light reduces germination efficiencies or cotyledon greening of Col-0 to a greater extent

### FIGURE 9 | Continued

under long-day conditions (A–C) and under continuous light (D). (A) Seed germination was investigated on MS medium in the absence and presence of 1 µM ABA. Radicle emergence was scored after indicated time points. (B) Phenotypes of 7-day-old seedlings grown on MS medium in the presence of 1 µM ABA. (C) The ratio of cotyledon greening was determined after 6 days. (D) Phenotypes of 7-day-old seedlings grown on MS medium in the presence of 1 µM ABA under continuous light. The pictures of the lower row are magnifications of the pictures above them. The data in (A,C) represent mean values ± SD of three independent experiments, each performed with at least 100 seeds per treatment and genotype. Statistically significant differences (t-test; p < 0.05) between WT (Col-0 or WS) and corresponding mutant lines are indicated by an asterisk (black for Col-0; red for WS).

(Reyes and Chua, 2007; Chen et al., 2008; Hwang et al., 2015) compared to long-day conditions (**Figure 7**; Hu et al., 2013). To tackle the oe-mTERF10 ABA phenotype, we grew Col-0, mterf mutant lines and oe-mTERF10 lines in continuous light and a temperature of 20◦C on MS medium (control) and MS medium supplemented with 1 µM ABA, and the phenotypes were scored after 7 days (**Figure 9D**). Because lines with a WS background were already very sensitive to ABA in long-day conditions, these lines were omitted. Indeed, Col-0 and mterf mutant seedlings displayed even shorter roots compared to long-day conditions and the cotyledons that had started to emerge did not turn green to this time point (**Figure 9D**). In contrast, abi4-1 seedlings had longer roots and fully expanded green cotyledons. All mutant lines with reduced MTERF10, -11, or -12 transcript levels showed the same behavior as Col-0. However, also in continuous light the phenotype of oe-mTERF10 lines was comparable to that of the abi4-1 mutant, with longer roots and expanded green cotyledons (**Figure 9D**).

In sum, these results indicate that lower or higher MTERF11 levels result both in increased ABA sensitivity. But, strikingly, higher MTERF10 levels are associated with decreased sensitivity to ABA, which might in turn be linked to the higher salt tolerance of oe-mTERF10 lines.

### DISCUSSION

Arabidopsis thaliana contains 35 mTERF proteins, of which seven have been investigated in more detail (reviewed in: (Kleine and Leister, 2015; Quesada, 2016)). Twenty-six of the 35 mTERFs have been sorted into groups based on their expression profiles and co-expression behavior (Kleine, 2012). The mTERF proteins that have been investigated in more detail are members of the "chloroplast" cluster (mTERF1, -4, -5, -6, and -9; the cluster itself contains 9 members) and the "mitochondrial" cluster (mTERF15 and -18; this cluster contains 7 members). In this study, we added to the inventory of characterized mTERFs and investigated all members of the "chloroplast-associated" cluster (mTERF10, -11, and -12).

In a fluorescence microscopy study of mTERF-GFP fusion proteins, 16 mTERFs were shown to be targeted to mitochondria, 11 to chloroplasts and one to the nucleus/cytosol (Babiychuk et al., 2011). That study revealed localization of mTERF4 to chloroplasts and mTERF6 to mitochondria. Later studies demonstrated that mTERF4 (Quesada et al., 2011) and mTERF6 (Romani et al., 2015) are in fact found in both mitochondria and chloroplasts. The mTERF proteins in the chloroplast-associated group were also assigned to the chloroplasts in the largescale study cited above (Babiychuk et al., 2011). However, to confirm or extend these data, and also to define the subchloroplast localization of mTERF10, -11, and -12, we transiently transformed A. thaliana protoplasts with GFP fusion proteins. By co-transformation with a RAP-RFP fusion protein, which is a marker for the chloroplast nucleoid (Kleinknecht et al., 2014), we show that all members of the chloroplast-associated cluster are localized to chloroplast nucleoids (see **Figure 1**). Also the maize homologs of Arabidopsis mTERFs-2, -3, -4, -5, -7, -9, -16, and -27 were identified in enriched maize nucleoids (Majeran et al., 2012) and Arabidopsis mTERF8 was found in preparations of the plastid transcriptionally active chromosome (pTAC; Pfalz et al., 2006) which is related to the nucleoid (Pfalz and Pfannschmidt, 2013). The nucleoid houses proteins that are associated with DNA organization, replication and repair, and furthermore, are involved in transcription, and processing, splicing and editing of transcripts, suggesting that mTERFs participate in PGE (Majeran et al., 2012). In fact, the three plant mTERF proteins whose molecular functions are known do participate in PGE: mTERF4 is involved in chloroplast group II intron splicing (Babiychuk et al., 2011; Hammani and Barkan, 2014), mTERF6 promotes maturation of a chloroplast tRNA (Romani et al., 2015) and in mterf15 mutants intron splicing of mitochondrial nad2 transcripts is perturbed (Hsu et al., 2014). Because levels of 16 and 23S rRNAs, and thus chloroplast protein synthesis, are reduced in the soldat10 mutant (Meskauskiene et al., 2009), it can be assumed that the mTERF1/SOLDAT10 protein is likewise involved in PGE.

Most of the previously characterized mterf mutants show phenotypes under normal growth conditions. Inactivation of mTERF1 (Meskauskiene et al., 2009) or mTERF4 (Babiychuk et al., 2011) is embryo lethal, the mterf6 and mterf15 knock-out mutants are seedling lethal (Romani et al., 2015) and retarded in growth and development (Hsu et al., 2014), respectively, and mda1 (mterf5) and mterf9 mutants are small and pale (Robles et al., 2012, 2015). We were unable to discern any phenotypic alterations in MTERF12 RNAi lines, either under normal or challenging growth conditions. In fact, mTERF12 might not be a bona fide mTERF protein, because an analysis with the SMART tool fails to identify any mTERF domain in mTERF12 (see above). On the other hand, MTERF12 is expressed at moderate to high levels in many developmental stages and organs (Kleine, 2012), and the mTERF12-eGFP fusion protein is localized to nucleoids (see **Figure 1B**). Therefore, while mTERF12 might not belong to the eponymous family, it may nevertheless be involved in PGE. Moreover, the residual amount of MTERF12 (32% of WT transcript levels) present in mterf12i-1 (see **Figure 3A**) may suffice to maintain a WT-like phenotype under all the conditions examined here, or alternatively we may not have hit upon the conditions required to provoke an abnormal phenotype in mterf12i lines. The latter possibility appears to be the more likely. For MTERF12 mRNA levels are highest in pollen (Wang et al., 2008; Kleine, 2012) and the most pronounced change in MTERF12 transcript level occurs in response to supplementation with nitrate (see **Supplementary Figure S3**), an intervention to which mterf12i lines were not subjected. Moreover, functional redundancy cannot be completely ruled out, although none of the three genes of interest originated from a duplication event (Kleine, 2012) and our phylogenetic tree (see **Figure 4**) and protein sequence comparisons (see above) do not strongly support this idea.

In addition to their pale color and growth-restricted phenotype, the mda1 (mterf5) and mterf9 mutants are less susceptible to salt and osmotic stresses, perhaps caused by reduced sensitivity to ABA (Robles et al., 2012, 2015). Notably, acclimation outputs are also altered by impairments in several other mTERF proteins (Meskauskiene et al., 2009; Quesada et al., 2011; Kim et al., 2012). Indeed, the emerging role of A. thaliana and maize mTERFs in acclimation and stress responses has already been noted and discussed (Zhao et al., 2014; Kleine and Leister, 2015; Quesada, 2016). This notion is especially of importance for crop plants, because plant development and growth is reduced in challenging growth conditions, leading finally to reduced yield. For this reason, several strategies have been tried to produce abiotic stress tolerance crop plants (Sah et al., 2016). With the aim to find a starting point to investigate stress tolerance in cotton, the response of cotton to abiotic stress treatments was studied with a cDNA library derived from samples treated with different stress conditions. Indeed, many transcripts for known stress-related genes, transcription factors and also mTERFs were enriched in this library (Zhou et al., 2016). Moreover, investigation of transcript level changes of six maize MTERF genes (maize MTERF2, -5, -11, -12, - 13, and -28) in response to salt, ABA and NAA treatments showed an upregulation of MTERF28 transcripts in all tested stress conditions, while MTERF12 transcript levels were induced nearly 2-fold after salt stress treatment. This suggests that of these tested mTERFs, maize mTERF28 is the strongest candidate participating in all tested stress responses, while mTERF12 might be especially involved in salt stress responses (Zhao et al., 2014). Our results show that in contrast to the strong mterf mutant phenotypes which point to essential functions of several mTERFs (Meskauskiene et al., 2009; Babiychuk et al., 2011; Romani et al., 2015), lines with altered MTERF10 or MTERF11 levels show only conditional phenotypes, which become manifest under adverse growth conditions (see **Figures 7**, **9**). Strikingly, under continuous light, lower MTERF10 levels are associated with reductions in salt tolerance, while oe-mTERF10 lines are more tolerant of salt and ABA than wild-type plants. The altered responsiveness to ABA of oge and also plastid signaling mutants has been noted before. For example, the "mitochondrial PPR protein PENTATRICOPEPTIDE REPEAT PROTEIN FOR GERMINATION ON NaCl" (PGN; Laluk et al., 2011), the tetrapyrrole biosynthesis proteins GUN4 and GUN5 (Voigt et al., 2010) and the plastid-targeted PPR protein GUN1 (Cottage et al., 2010) alter responses to ABA. Notably, gun1 mutants show only

subtle growth phenotypes, but GUN1 is an important integrator of plastid signals (Koussevitzky et al., 2007). Like mTERF proteins, PPR proteins are typically targeted to chloroplasts or mitochondria, and alter expression of transcripts by influencing editing, turnover, processing or translation (Barkan and Small, 2014). With more than 400 members, the PPR protein family is one of the largest in land plants (Barkan and Small, 2014) and far exceeds the mTERF family in size. The enlargement of the plant PPR family has been linked to the evolution of a complex organellar gene expression system that is characteristic for plant organelles (Barkan and Small, 2014), and this is likely to be true of the mTERF protein family also (Kleine, 2012). Moreover, and in contrast to animals, plants are sessile organisms that are exposed to environmental changes and stresses. During evolution, the expansion and functional diversification of protein families has helped plants to successfully adapt to or tolerate different environmental stresses (Quesada, 2016). The mTERF family is a good case study for this phenomenon. With the characterization of an increasing number of plant mTERF proteins, it is becoming evident that they play a wide range of roles in mediating tolerance and acclimation to different abiotic stresses.

### AUTHOR CONTRIBUTORS

Research was designed by TK. Research was performed by DX and TK. The manuscript was prepared by DX, DL, and TK.

## FUNDING

This work was supported by the Deutsche Forschungsgemeinschaft [KL 2362/1-1 to TK, and TRR175 to DL (project C05) and TK (project C01)].

## ACKNOWLEDGMENTS

We thank Elisabeth Gerick for technical assistance and Jörg Nickelsen for providing the RAP construct.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2017. 01213/full#supplementary-material

Supplementary Figure S1 | Quantification of changes in MTERF10 mRNA expression in response to perturbations as determined with the Genevestigator Perturbations Tool. The tool was employed on all deposited A. thaliana ATH1 arrays together with a 2-fold change filter and a p-value of < 0.05.

Supplementary Figure S2 | Quantification of changes in MTERF11 mRNA expression in response to perturbations as determined with the Genevestigator Perturbations Tool. The tool was employed on all deposited A. thaliana ATH1 arrays together with a 2-fold change filter and a p-value of < 0.05.

Supplementary Figure S3 | Quantification of changes in MTERF12 mRNA expression in response to perturbations as determined with the Genevestigator Perturbations Tool. The tool was employed on all deposited A. thaliana ATH1 arrays together with a 2-fold change filter and a p-value of < 0.05.

Supplementary Figure S4 | Behavior of wild-type (Col-0) and mterf10, -11 and -12 mutant plants under moderate light stress of 400 µmol photons

m−<sup>2</sup> s −1 . (A) To score the phenotypes under moderate heat stress, plants were first grown for 3 weeks under normal growth conditions (100 µmol photons m−<sup>2</sup> s −1 , 22◦C) and then exposed to 400 µmol photons m−<sup>2</sup> s −1 for 7 days. (B) The maximum quantum yield of PSII (Fv/Fm) of Col-0 and

### REFERENCES


mterf mutant plants was determined after the indicated periods of exposure to a fluence of 400 µmol photons m−<sup>2</sup> s −1 . The data are shown as mean values ± SD from 8 to 10 different leaves.

Supplementary Table S1 | Primers used in this study.


Quesada, V. (2016). The roles of mitochondrial transcription termination factors (MTERFs) in plants. Physiol. Plant. 157, 389–399. doi: 10.1111/ppl.12416


at the light-strand promoter but is dispensable for rRNA gene transcription regulation. Cell Metab. 17, 618–626. doi: 10.1016/j.cmet.2013.03.006


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Xu, Leister and Kleine. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Import of Soluble Proteins into Chloroplasts and Potential Regulatory Mechanisms

Inga Sjuts<sup>1</sup> , Jürgen Soll1,2 and Bettina Bölter1,2 \*

<sup>1</sup> Department Biologie I-Botanik, Ludwig-Maximilians-Universität, Planegg-Martinsried, Germany, <sup>2</sup> Munich Center for Integrated Protein Science CiPSM, Ludwig-Maximilians-Universität, Munich, Germany

#### Edited by:

Fiammetta Alagna, Research Council for Agriculture and Agricultural Economy (CREA), Italy

#### Reviewed by:

Matthew D. Smith, Wilfrid Laurier University, Canada Vivien Rolland, Commonwealth Scientific and Industrial Research Organisation, Australia Felix Kessler, University of Neuchâtel, Switzerland

### \*Correspondence:

Bettina Bölter boelter@bio.lmu.de

#### Specialty section:

This article was submitted to Plant Cell Biology, a section of the journal Frontiers in Plant Science

Received: 28 October 2016 Accepted: 26 January 2017 Published: 08 February 2017

#### Citation:

Sjuts I, Soll J and Bölter B (2017) Import of Soluble Proteins into Chloroplasts and Potential Regulatory Mechanisms. Front. Plant Sci. 8:168. doi: 10.3389/fpls.2017.00168 Chloroplasts originated from an endosymbiotic event in which a free-living cyanobacterium was engulfed by an ancestral eukaryotic host. During evolution the majority of the chloroplast genetic information was transferred to the host cell nucleus. As a consequence, proteins formerly encoded by the chloroplast genome are now translated in the cytosol and must be subsequently imported into the chloroplast. This process involves three steps: (i) cytosolic sorting procedures, (ii) binding to the designated receptor-equipped target organelle and (iii) the consecutive translocation process. During import, proteins have to overcome the two barriers of the chloroplast envelope, namely the outer envelope membrane (OEM) and the inner envelope membrane (IEM). In the majority of cases, this is facilitated by two distinct multiprotein complexes, located in the OEM and IEM, respectively, designated TOC and TIC. Plants are constantly exposed to fluctuating environmental conditions such as temperature and light and must therefore regulate protein composition within the chloroplast to ensure optimal functioning of elementary processes such as photosynthesis. In this review we will discuss the recent models of each individual import stage with regard to short-term strategies that plants might use to potentially acclimate to changes in their environmental conditions and preserve the chloroplast protein homeostasis.

Keywords: chloroplast, protein import, TOC, TIC, plastid proteostasis, acclimation

## INTRODUCTION

Chloroplasts are unique photosynthetic organelles that evolved through an endosymbiotic event ∼1.5 billion years ago. A formerly free-living cyanobacterium was engulfed via phagocytosis by an ancestral eukaryotic host that already contained mitochondria (Gould et al., 2008). During evolution, a dramatic reduction in the bacterial endosymbiont genome size occurred, since 95% of the genes encoding the ∼3000 proteins acting in the chloroplasts were transferred to the host nucleus that attained control over its new organelle. The plastid genome encodes the residual ∼100 genes (Sugiura, 1989; Martin et al., 2002; Timmis et al., 2004). As a consequence, nuclear-encoded chloroplast proteins that were originally encoded on the endosymbiont genome are now translated in the cytosol and are post-translationally translocated into the allocated organelle (Leister, 2003). This translocation process requires a first-sorting event of the so-called preproteins. According

to their chloroplast-specific targeting peptide, which is called chloroplast transit peptide (cTP), the preproteins are targeted to the receptor-equipped destination organelle. Previously, this term has been used to define both the (1) N-terminal peptide which is cleaved off in the stroma upon import and (2) the sequence which is necessary and sufficient for import of a cargo protein into the chloroplast. However, these peptides differ from each other, as the sequence of (1) is determined by the processing site and does not contain parts of the mature protein, whereas (2) could also include domains from the mature protein. In order to avoid confusion, Rolland et al tried to find a suitable nomenclature for this issue. The term cTP refers to the sequence of the preprotein which is required for chloroplast targeting and cleaved off upon import. The cTP is determined by the processing site and is not part of the mature protein. In contrast, the sequence which is necessary and sufficient for import of a cargo protein into the chloroplast is called transit peptide. This transit peptide includes the cTP and possibly part of the mature protein (Comai et al., 1988; Bionda et al., 2010; Rolland et al., 2016). During targeting, preproteins can interact with different cytosolic chaperones that enable the cell to keep the preproteins in an unfolded, and hence import-competent, state. After recruiting the chaperoned complexes to the chloroplast outer envelope membrane, translocation is initiated. In the majority of cases, import across the outer and inner envelope membrane of chloroplasts is facilitated by two distinct translocation complexes, called TOC (translocon on the outer chloroplast membrane) and TIC (translocon on the inner chloroplast membrane), respectively. Once inside the stroma, a stromal processing peptidase (SPP) cleaves off the cTP, and the remaining mature protein undergoes folding and insertion or further direction to intraorganeller targets, again with the guidance of stromal chaperones (Richter and Lamppa, 1998; Richter et al., 2005).

It has been long known that import activity correlates with protein demand during plastid development of a plant life (Dahlin and Cline, 1991). In young and fast dividing tissues, the protein demand is especially high, in comparison to adult and non-dividing cell parts. It has been shown that protein import into plastids is developmentally regulated (Li and Teng, 2013). However, since plants are sessile organisms, even mature tissues of a plant are constantly exposed to fluctuating environmental conditions such as temperature and light, and plants must therefore regulate their protein content within the chloroplast to ensure optimal functioning of processes such as photosynthesis. Specifically, the photosynthesis rate depends on different intensities of light and temperature, hence all subunits of the involved complexes have to be produced, imported and assembled according to the current demand.

Several studies noticed that the actual chloroplast proteome is indeed influenced by short-term applications of varying temperature or light conditions, meaning the plant is effectively acclimating upon external stimuli (Dutta et al., 2009; Grimaud et al., 2013). In contrast to adaptation, during which a trait evolves over a longer period of time by means of natural selection, in our understanding acclimation refers to an environmentally inducible and mostly even reversible event which occurs within the organism's lifetime. Several upstream mechanisms exist, such as changes in the transcription rate of preproteins or involved import receptors upon stress applications. One mentionable example is the upregulation of the TOC GTPase genes upon salt stress in tomato seedlings (Yan et al., 2014). However, these transcriptional mechanisms will not be part of this review; instead, the chloroplast protein import process itself is one advisable target to be highly regulated at different stages, thus leading to a dynamic acclimation of import activity. This acclimation can be achieved by means of posttranslational mechanisms such as reversible phosphorylation or oxidation/reduction of both to-be imported and import-related proteins.

Here, we review the individual steps involved in protein translocation into chloroplasts and touch on regulation mechanisms that plants might use to modulate protein import. It is worth mentioning that our understanding of import regulation is still developing. Therefore, we have tried to summarize what is known so far and what the available data from different research groups might mean concerning regulatory mechanisms. These overall speculations might contribute to our current understanding of how plants potentially acclimate to external stimuli by fine-tuning their organellar protein import.

### CYTOSOLIC SORTING OF PREPROTEINS AND TARGETING TO THE ORGANELLE – THE ROLE OF REVERSIBLE PHOSPHORYLATION

After completion of translation on cytoplasmic ribosomes, the initial step of protein import is the accurate targeting of these newly synthesized preproteins. To avoid mistargeting, chloroplast-destined preproteins harbor an N-terminal cTP that specifically targets them to the chloroplast outer membrane (Bruce, 2001). Unexpectedly, conserved characteristics specific to chloroplast proteins across plant species are missing and the sequences of cTPs are highly heterogeneous in their length and properties. They merely display an overall positive net charge, resulting from the lack of acidic amino acids (Bruce, 2001). Regarding the fact that mitochondrial proteins have specific and conserved features within their N-terminal targeting sequence across plant species, the lack of such a consensus sequence for chloroplast-targeted proteins is striking, thus rendering the question of how specificity for the chloroplast is achieved and mistargeting between these organelles is avoided. One potential hypothesis for the heterogeneity could be different preferences of the preproteins for plastid types, which is determined by distinct cTP features (Li and Teng, 2013).

To sustain import competency by keeping preproteins in an unfolded structure, cytosolic chaperones are involved. Up to now, the most prominent chaperone thought to facilitate appropriate

recruiting of preproteins is Hsp70. Both cTPs and the mature part of preproteins have been shown to interact directly with this chaperone, and import activity is clearly stimulated in the presence of Hsp70 (Rial et al., 2000).

Apart from Hsp70, another component has been identified in cytosolic preprotein targeting: a 14-3-3 protein preferentially binds to phosphorylated serines or threonines within the cTP, which in association with the chaperone Hsp70 leads to increased import efficiency of preproteins. This assembly has been designated the cytosolic guidance complex (May and Soll, 2000) (**Figure 1**). Phosphorylation is mediated by the recently identified STY kinases 7, 18, and 46; a knockout of two and concurrent knockdown of the third kinase led to severe phenotypes in chloroplast biogenesis during greening (Lamberti et al., 2011). However, it seemed that dephosphorylation plays a more crucial role in the actual import process than phosphorylation. It could be shown that under the applied conditions – removal of the phosphorylation site within the binding motif of the cTP for 14-3-3 proteins – the kinetics, rather than the fidelity, of targeting to chloroplasts was impaired (May and Soll, 2000; Nakrieko et al., 2004). In contrast, phosphorylated precursors, or those containing a glutamic acid residue instead to mimic phosphorylation, are only imported very slowly (Waegemann and Soll, 1996). In vivo studies showed that an Arabidopsis mutant which mimicked the phosphorylated serine in the cTP of the photosynthetic precursor pHcf136 resulted in reduced import activity, and hence impaired photosystem II assembly, most prominent in cotyledons (Nickel et al., 2015). This is probably due to the impossibility of dephosphorylation occurring within the cTP and clearly demonstrates that import and assembly of photosynthetic proteins is highly dependent on a proper phosphorylation/dephosphorylation cycle prior to translocation. Once this process cannot be completed, the chloroplast protein homeostasis is misbalanced.

Like Hsp70, the chaperone Hsp90 is able to bind to both the cTP and mature region of a different subset of preproteins. Its presence alone stimulates protein import into isolated chloroplasts (Qbadou et al., 2006; Fellerer et al., 2011). In contrast to the Hsp70/14-3-3 guidance complex, Hsp90-bound preprotein favors a distinct docking station at the OEM, which will be defined below.

As neither the guidance complex nor the phosphorylation event is essential for successful import, it is highly tempting to speculate that under specific conditions phosphorylation has a regulatory function rather than an essential role in protein import. As phosphorylation is generally a fast response, one can assume that different external stimuli trigger the phosphorylation to regulate protein import. Independent from protein import, this has been shown not only for light-dependent phosphorylation in photosynthetic reactions, but also as a general response to different stress stimuli (Grieco et al., 2016).

It would be interesting to know if the in vivo phosphorylation/dephosphorylation circuit of preproteins is enhanced or reduced under stress conditions such as high light treatment, and to define the influence of this regulation mechanism on protein import. Furthermore, whether this effect

would be due to either an enhanced activation of the mentioned STY kinases or due to the inhibition of the yet unknown phosphatase remains an interesting question to address.

## CROSSING THE OUTER ENVELOPE MEMBRANE VIA THE TOC COMPLEX

After synthesis and sorting in the cytosol, the preproteins are recognized at the OEM. This is mainly mediated by the two GTP-dependent receptor proteins Toc34 and Toc159 (Kessler and Schnell, 2009). Both proteins are anchored C-terminally in the OEM and expose their GTP-binding domains toward the cytosol, in consistency with their role as preprotein receptors. Together with a third protein, Toc75, which is deeply embedded in the lipid bilayer and forms the protein conducting channel (Hinnah et al., 1997), they build up a stable complex, resulting in a heterotrimeric TOC core complex (**Figure 1**). Determination of the apparent mass of 500 kDa of the pea multiprotein complex leads to a stoichiometry of 1:4:4 of Toc159/Toc34/Toc75 (Schleiff et al., 2003). Both receptors belong to a plant-specific family of eukaryotic-originated GTPases, sharing some general features. Toc159 is a tripartite protein consisting of three functional domains: an intrinsically disordered acidic domain (A-domain), the GTPase domain (G-domain) and the membrane anchor domain (M-Domain with a mass of ∼54 kDa) (Bölter et al.,

1998a; Chen et al., 2000; Richardson et al., 2009). Toc34 contains a cytosolic GTPase domain and is anchored into the OEM by a single transmembrane domain. Both proteins Toc34 and Toc159 bind to distinct regions of the N-terminal cTP, hence they could act simultaneously in receiving preproteins (Becker et al., 2004).

The GTPase activity plays a central role in preprotein recognition and delivery, as non-hydrolyzable GTP analogs inhibit preprotein binding and translocation (Young et al., 1999). Interestingly each individual GTPase domain is dispensable for the plant (Agne et al., 2009; Aronsson et al., 2010), however, a viable plant lacking both domains from both receptors could not yet be isolated. The minimal structure required for sufficient assembly of the TOC complex and to support protein import is the M-domain of Toc159, which can partially complement the loss of Toc159 in ppi2 mutant plants (Lee et al., 2003).

Toc34 is believed to exist as a homodimer in its GDPbound state, which exhibits a preprotein-binding site in its GTPase domain (Sun et al., 2002). Upon preprotein delivery, GTPase activity is stimulated and exchanges GDP to GTP. Toc34 in its GTP-bound state binds preproteins with high affinity, which triggers not only the disruption of the Toc34 dimer but also promotes heterodimerization of Toc34 and Toc159. This GTP-heterodimer-complex is now referred to as the active TOC complex (Becker et al., 2004). GTP hydrolysis results in reduced affinity toward the preprotein, the subsequent transfer of the preprotein into the Toc75 channel and the initiation of membrane translocation (Oreb et al., 2007). Taken together, the hypothesized model clearly demonstrates that the receptors are working as GTP/GDP-regulated switches to control preprotein binding and delivery. However, there are still missing factors, such as the GTPase-activating protein or GTP-exchange factor, although it could be shown that peptides from cTPs can stimulate GTPase activity (Jelic et al., 2003).

In Arabidopsis, different homologs of the TOC receptors exist, which enhances complexity and specificity toward binding proteins. The Toc159 family consist of four genes, each of them differentially participating in chloroplast biogenesis. These are atToc159, atToc132, atToc120 and atToc90, which show high similarity in their G and M domains, but a high variation in sequence and length of the dynamic A-domain (Bauer et al., 2000; Kubis et al., 2004).

The most abundant isoform is atToc159, consequently the knockout toc159 (ppi2) mutant shows an albino phenotype and is seedling lethal, but can grow hetero-autotrophically (Bauer et al., 2000; Bischof et al., 2011). The latter, and the fact that atToc159 exhibits high expression levels in juvenile developmental stages, led to the suggestion that atToc159 constitutes the primary receptor for photosynthetic precursor proteins, which will be discussed below (Bauer et al., 2000). AtToc90 can complement the albino phenotype of ppi2 and restores photoautotrophic growth, indicating that atToc90 has a similar function to atToc159 (Infanger et al., 2011). Based on expression pattern and the ability to rescue the toc159 mutant phenotype, the different TOC receptors are classified in two groups: the abovementioned group of atToc159 and atToc90, and a second group consisting of atToc132 and atToc120. AtToc132 and atToc120 are expressed at similar levels throughout all tissues and are functionally exchangeable (Ivanova et al., 2004; Kubis et al., 2004). However, atToc120 cannot rescue the phenotype of toc159, clearly emphasizing a distinct specificity toward preproteins.

The Arabidopsis Toc34 family comprises atToc34 and atToc33, which likewise display differential developmental expression profiles. AtToc33 is highly expressed in juvenile, photosynthetic-active tissues, whereas atToc34 is expressed at low levels throughout all developmental stages and all organs. In line with this expression profile, the toc33 (ppi1) mutant showed a pale phenotype during early development, but reached near-WT appearance after 2 weeks of growth. AtToc33 and atToc34 functionally overlap. Different observations led to this conclusion. First, both proteins show 65% sequence similarity; secondly, a small fraction of atToc33 co-immunoprecipitated with atToc120/atToc132, which was originally shown only for atToc34; thirdly, the double knockout of atToc33 and atToc34 is embryo lethal; and last and most critically, atToc34 can complement the ppi1 phenotype (Jarvis et al., 1998; Ivanova et al., 2004; Kubis et al., 2004).

Different studies led to the overall assumption that various isoforms of the GTPases associate with distinct TOC complexes and may prefer a particular set of precursors. It was suggested that atToc159/atToc90 bind to atToc33, whereas atToc120 and/or atToc132 form a complex together with atToc34 (Ivanova et al., 2004). Originally the idea was favored that the various TOC complexes represent distinct pathways for incoming preproteins. It was stated that the complex consisting of the most abundant isoforms atToc159 and atToc33 preferentially imports highly demanded photosynthetic preproteins, whereas the other TOC isoforms form a translocation complex with specificity toward housekeeping proteins (Ivanova et al., 2004; Kubis et al., 2004; Smith et al., 2004). However, this oversimplified model has been rejected due to a large-scale proteomic and transcriptomic approach by Bischof et al. (2011), in which they identified an import defect for different functional subsets of preproteins in ppi2 protoplasts. Similar to this observation, equal numbers of photosynthetic and non-photosynthetic preproteins were identified to interact with both atToc159 and atToc132 (Dutta et al., 2014). Nevertheless, the distinct preferential import pathways could be a subtle hint for selectivity of target preproteins, possibly in different developmental stages or under diverse external environmental conditions. As the Toc34 isoforms are functionally interchangeable, preprotein selectivity could be mediated by the Toc159 family. Recent hints are pointing toward a specificity-conferring role of the variable A-domains of the Toc159 receptor family (Inoue et al., 2010).

A third component was identified to assist in receiving preproteins, named Toc64. Its potential role in protein import has been concluded from its ability to bind a precursor protein and the transient association with the other TOC components (Sohrt and Soll, 2000). In contrast to the above-mentioned receptor proteins, Toc64 serves as an initial docking station for Hsp90-bound preproteins und subsequently delivers these

preproteins to Toc34 (Qbadou et al., 2006). Toc64 harbors three cytosolic tetratricopeptide repeat (TPR) domains, mediating the interaction with Hsp90 (**Figure 1**). This is a typical feature of proteins interacting with Hsp70/90-associated proteins (Young et al., 2003). The same holds true for a plant ER receptor TPR7 (Schweiger et al., 2012) and interestingly, a Toc64 homolog, namely OM64, was found in plant mitochondria, replacing the mitochondrial TOM70 present in mammals and fungi but absent in plants. Instead, the protein OM64 with a C-terminal TPR domain serves as a receptor for mitochondrial-destined proteins (Chew et al., 2004). Although in vitro a strong interaction between Hsp90 and Toc64 could be measured with a K<sup>D</sup> of 2.4–15.5 µm (Schweiger et al., 2012) the essentiality of these TPR proteins in vivo is still under debate. Since chloroplasts lacking Toc64 sustain their import capacity, it is feasible that this docking protein rather constitutes more an additional regulatory component to the general TOC receptor complex than being an essential constituent. However, it could be shown that atToc33 and Toc64 cooperate in preprotein import, hence it is reasonable to say that atToc33 can overcome the loss of Toc64 function as preproteins are still recognized (Sommer et al., 2013), while only chaperone binding is lost.

After the preprotein has been delivered to the receptor proteins, it has to be translocated through the membrane. The preprotein-conducting channel in the OEM is represented by the beta barrel protein Toc75 (Schnell et al., 1994). The essential nature of Toc75 is demonstrated by its gene being a single copy conserved throughout all plant lineages and the embryo lethality of knockout lines (Jackson-Constan and Keegstra, 2001). The protein belongs to the Omp85 superfamily, which is exclusively found in gram-negative bacteria, mitochondria and plastids (Bölter et al., 1998b). Typically for this family, the structure of Toc75 exhibits two features: 16-18 arranged beta strands forming the C-terminal beta domain, and several POTRA domains at its N-terminus (Clantin et al., 2007). Irrespective of the fact that POTRA domains are required for Toc75 function (Paila et al., 2016), the orientation and thus exact molecular function of these POTRA domains remain a matter of debate. On the one hand, it is assumed that these domains are facing the cytosolic side of the OEM, assisting in preprotein interaction (Sommer et al., 2011). However, a recent study proposed a localization of the POTRA domains in the intermembrane space (Chen et al., 2016).

In vitro analyses showed preprotein binding during import and the import process itself being inhibited with Toc75 antibodies (Tranel et al., 1995). Electrophysiological analyses revealed that reconstituted Toc75 in lipid bilayers forms a voltage-gated channel with a pore size of 14Å at its narrowest part (Hinnah et al., 2002). In contrast to the other TOC components, Toc75 harbors an N-terminal bipartite transit peptide. One part directs the protein into the stroma where the SPP cleaves off this portion once the extreme N-terminus reaches the stroma, whereas the second cleavage site is processed by a plastidic type I signal peptidase (Plsp1), which is localized to the IEM (Inoue et al., 2005).

## CROSSING THE INTERMEMBRANE SPACE AND INNER ENVELOPE MEMBRANE VIA THE TIC COMPLEX

Successful import requires not only the interaction between preproteins and outer membrane receptors, but also the formation of super complexes between the translocons of both OEM and IEM via contact sites that enable the preprotein to pass through both membranes simultaneously (Schnell and Blobel, 1993). Both complexes are facing the intermembrane space, thus some proteins localized in this compartment have to be involved in the import process. However, only limited knowledge about import-related factors of the intermembrane space is available. Presently, the only member identified in this compartment to be involved in protein translocation is the soluble protein Tic22. Tic22 has been shown to interact with preproteins during protein import (Kouranov et al., 1998). Structural and functional studies led to the hypothesis that Tic22 is working as a molecular chaperone, as Arabidopsis mutants lacking Tic22 showed growth and biogenesis defect and a decreased import activity (Kasmati et al., 2013; Rudolf et al., 2013). One potential role for Tic22 would be, like the cytosolic counterparts, to ensure proper targeting and prevent misfolding during the transfer between TOC and TIC. However, this role has not been confirmed yet.

The TIC counterpart of the TOC core channel Toc75 is Tic110. Tic110 was the first TIC component described Schnell et al. (1994) and is the second most abundant protein in the IEM (Lübeck et al., 1996). It was found in a supercomplex associated with TOC components and incoming preproteins, suggesting a functional role as the central part of the IEM translocon (Lübeck et al., 1996).

Reconstitution of a Tic110-protein lacking its two N-terminal hydrophobic transmembrane stretches (pea sequence: aa91-966, 1N-110) resulted in a cation-selective channel with a diameter of 1.7 nm, which is similar to the diameter of the channel of Toc75 and hence sufficient for preprotein threading (Heins et al., 2002; Balsera et al., 2009) (**Figure 2**). However, two controversial models concerning the topology and function of Tic110 still persist. Undoubtedly and universally accepted is the fact that the 110-kDa protein is anchored into the membrane by its two N-terminal, highly hydrophobic helices (Inaba et al., 2003; Balsera et al., 2009). In our current topological model, we can combine the essential functions of Tic110, which has been under discussion for a long time. On the one hand, Tic110 assembles into its channel-like structure via its four amphipathic helices, substantiating its function as the main translocation pore. The four membrane-spanning helices consequently lead to the formation of two loops that are extended into the intermembrane space, which could be confirmed by limited proteolysis experiments (Lübeck et al., 1996; Balsera et al., 2009). On the other hand, a large part of the C-terminus is protruding into the stroma and thus could fulfill the additional function of Tic110 acting as a scaffold for chaperones and co-chaperones (Inaba et al., 2005). The crystal structure of a Cyanidioschyzon merolae Tic110 version, which consists of the C-terminus including only the last amphipathic helix, is proposed

to be too flattened and elongated to form a channel protein (Tsai et al., 2013). However, as it is unlikely that such a shortened protein can fold into its native conformational structure, it is still reasonable to assume that the full-length Tic110 protein is able to build the channel protein via its amphipathic, membranespanning helices.

Like Toc75, Tic110 is encoded by a single gene and constitutively expressed in all tissues. Homozygous T-DNA insertion lines are embryolethal, and heterozygous plants already exhibit a clear growth and greening defect, clearly emphasizing the necessity of Tic110 in chloroplast biogenesis and overall plant viability (Kovacheva et al., 2005). Import of Tic110 is achieved by targeting the protein into the stroma and after cleavage of the cTP, Tic110 is re-inserted into the lipid bilayer of the IEM (Vojta et al., 2007).

Using a cross-linking strategy, another TIC component could be directly associated to Tic110, named Tic40. Tic40 consists of a single transmembrane helix which anchors the protein at the IEM, resulting in a large stroma-facing, soluble domain. This C-terminal part harbors two Hip/Hop/Sti domains, building binding sites for Tic110 and the stromal Hsp70/93 chaperones. The main function of Tic40 is to co-chaperone the translocation process of incoming preproteins by coordinating Hsp93 activity (Chou et al., 2006) (**Figure 2**).

A further TIC component, named Tic20, was identified by its ability to covalently cross-link with a precursor protein en route to the chloroplast (Kouranov and Schnell, 1997; Kouranov et al., 1998). Structural prediction indicated three or four hydrophobic transmembrane domains (Kouranov et al., 1998). Tic20 is essential in Arabidopsis. Chloroplasts isolated from Tic20 antisense lines are impaired in preprotein import (Chen et al., 2002). In addition, early phylogenetic analysis indicated a relation of Tic20 with bacterial amino acid transporter and cyanobacterial proteins of unknown function suggesting a role as a translocation channel (Reumann and Keegstra, 1999). However, a latter study including many more genomes was unable to reproduce these claims (Gross and Bhattacharya, 2009). Nonetheless, the important role of Tic20 in chloroplast biogenesis is evident and it was proposed early on by Reumann et al. (2005) that Tic20 and Tic110 form independent preprotein translocation channels. Besides this circumstantial evidence for the notion, direct support comes from electrophysiological studies using either heterologously expressed and purified Tic20 (Kovács-Bogdán et al., 2011) or a 1MDa-complex from Arabidopsis, of which Tic20 is one constituent (Kikuchi et al., 2009, see below), which both showed the channel-forming capacity of the applied material. Using a cleavable proteinA-tagged variant of Tic20 expressed in transgenic Arabidopsis plants, the authors were able to purify the 1MDa complex via affinity purification. The obtained complex contained three other proteins in addition to Tic20: atTic56, atTic100 and atTic214 (Ycf1) (Kikuchi et al., 2013) (**Figure 2**).

Interestingly, Ycf1 is one of the last enigmatic open-reading frames of the chloroplast genome without an assigned function (Drescher et al., 2000). It is predicted to contain at least six transmembrane helices at its N-terminus (de Vries et al., 2015). AtTic56 and atTic100 are nuclear-encoded proteins, the first deeply embedded in the holo-complex without any predicted transmembrane domain, whereas the latter is supposed to associate with the complex on the intermembrane space site (Kikuchi et al., 2013). However, major questions came up concerning the exact physiological roles of the involved proteins. So far, for the potential involvement of Tic100, no data are available. However, for atTic56, a proteomic analysis showed that most of the chloroplast proteins are still imported into the organelle in atTic56 mutant plants, pointing toward a still functioning import machinery (Köhler et al., 2015). Furthermore, an alternative role independent from protein import for atTic56 was suggested, since Köhler et al. established a link between processing of plastid rRNA and the assembly of plastid ribosomes. They stated that a defect in plastid ribosome construction is responsible for the albino phenotype of atTic56-1 mutant plants, thus leading to a potential role of atTic56 in ribosome assembly and establishment of a functional plastid translation machinery (Köhler et al., 2016). Even more importantly, since Ycf1 is missing not only in all grasses but also in a variety of dicotyledonous plants, one can speculate about its overall significance in protein import. The critical question is: how do plants that are completely lacking this gene manage to retain their functional import machinery (de Vries et al., 2015)? Since Ycf1 is an essential protein in Arabidopsis, it is difficult to study protein import in knockout plants. Nonetheless, ecotypes of Arabidopsis can be grown on media containing spectinomycin, which is a specific inhibitor of plastid translation (Wirmer and Westhof, 2006). Under these conditions it could be shown that Ycf1 is truly absent in Arabidopsis plants, thus enabling to study its role in protein import (Bölter and Soll, 2016; Köhler et al., 2016). Presumably, the seed contains sufficient Ycf1 protein for the plants to germinate, and spectinomycininduced signaling leads to compensatory mechanisms that ensure

survival on the antibiotic. Interestingly, as these two studies show, precursor proteins that depend on the general protein import machinery are still efficiently imported into the plastids, thus excluding the role of Yfc1 as a constituent of the main protein channel. Furthermore, the nuclear-encoded Tic20 is also not detectable under spectinomycin treatment, implying a feedback mechanism between plastid and nucleus concerning the assembly of the 1MDa complex (Bölter and Soll, 2016). Instead of being a main translocation factor, Ycf1 could be involved in the assembly of a plastid fatty acid synthase (ACCase). Under spectinomycin, plants are also lacking the plastid-encoded subunit AccD but are able to complement for that loss by upregulating the expression and import of a nuclear-encoded and plastid-targeted protein (Acc2). This upregulation only appears if Ycf1 is strongly diminished, suggesting a functional role of Ycf1 in assembling the ACCase holoenzyme (Bölter and Soll, 2016). Recently, Ycf1 was shown to be a target of a nuclearencoded translational activator named PBR1, which is important for thylakoid biogenesis, suggesting it could play a role in this process (Yang et al., 2016). Although a potential role of Ycf1 in protein import cannot entirely be excluded, more research is needed to clarify its functional role(s).

Besides the discrepancies concerning the main translocation machinery, additional TIC components have been identified which are called the redox regulon. This regulon includes the proteins Tic55, Tic62, and Tic32 (Stengel et al., 2009). Tic55 is a Rieske protein, while both Tic62 and Tic32 are dehydrogenases. All proteins have been found in complexes containing Tic110; specifically, Tic32 shows a direct interaction with the N-terminus of Tic110 (Hörmann et al., 2004; Stengel et al., 2009). The role of these redox regulon members will be discussed in detail below.

### COMPLETION OF THE TRANSLOCATION PROCESS: THE STROMAL CHAPERONE SYSTEM

Upon reaching the stroma, the preprotein translocation proceeds by removing the cTP and subsequently folding into an active structure. Four distinct destinations for the imported proteins are possible: stroma, IEM, thylakoids and thylakoid lumen. The mature protein is either re-inserted into the IEM or, due to a bipartite transit peptide, directed to the thylakoids using different sorting mechanisms for further processing and assembly (Schünemann, 2007). The removal of the cTP is carried out by a soluble SPP which is essential for plants (Richter and Lamppa, 1998; Trösch and Jarvis, 2011). Import is an energy-consuming process resulting from nucleotide-hydrolysis. Although the TOC members are able to hydrolyze GTP, this provides only the minimal energy required for the irreversible initiation of protein import and is not the driving force for sufficient and complete import, so the energy must originate from a different source. It has been shown that the energy is provided in the form of ATP, which is hydrolyzed by stromal chaperones, leading to a sufficient motor activity for preprotein crossing of the OEM and IEM of the chloroplast (Pain and Blobel, 1987). Various chaperones have been determined as being involved in the folding of proteins and/or consuming the required energy via ATP hydrolysis, mainly the chloroplast Hsp70, Hsp90, Hsp93 and Cpn60 (Kessler and Blobel, 1996; Akita et al., 1997; Nielsen et al., 1997; Inoue et al., 2013). However, Cpn60, the homolog of bacterial GroEL, is most likely exclusively involved in protein folding and assembly of the newly imported mature proteins, especially Rubisco (Goloubinoff et al., 1989).

Hsp93 (bacterial ClpC) is a member of the Hsp100 family, which itself belongs to the broader AAA+ family (ATPases associated with various cellular activities) (Moore and Keegstra, 1993). Hsp100 proteins contain one or two AAA+ domains, and are typically arranged into a hexameric structure with a central pore which is sufficient for protein threading (Rosano et al., 2011). Arabidopsis features two genes encoding for the isoforms Hsp93-V and Hsp93-III. Beside the putative function of providing energy coming from ATP hydrolysis, Hsp93 has been shown to be a regulatory chaperone for the Clp protease system, thus functioning in quality control and potential degradation of the incoming preproteins (Kovacheva et al., 2005).

Originally, three chloroplast Hsp70 isoforms in pea were reported. Two of them are located in the stroma whereas one is supposed to reside in the intermembrane space (Ratnayake et al., 2008). However, in Arabidopsis the gene coding for the latter has not yet been identified, leaving doubts about the existence or identity of such an intermembrane-space chaperone. Arabidopsis double null mutants of the stromal Hsp70 isoforms are embryo lethal and single mutants already exhibit biogenesis and import defects (Su and Li, 2010).

CpHsp90 was identified in complexes containing import intermediates at late import stages that also contain Tic110 and Hsp93 (Inoue et al., 2013). A specific and reversible Hsp90 ATPase inhibitor arrests protein import in chloroplasts, whereas initial binding to the TOC complex is not impaired, clearly emphasizing a role of cpHsp90 in late import stages (Nakamoto et al., 2014).

Due to the complexity of the chaperone system in chloroplasts, there is an ongoing discussion about the specificity and importrelated function of each individual chaperone, resulting in different models. It is still not completely clear which protein is the potential candidate to constitute the main motor protein for providing the import energy. In mitochondria and ER, the responsible driving force is believed to come from ATP hydrolysis performed by Hsp70 chaperones which are located in the matrix and lumen, respectively (Park and Rapoport, 2012; Dudek et al., 2013). Thus, it was long thought that cpHsp70s are likewise the main motor in chloroplasts. In that context, it seems logic that the responsible ATPase interacts directly with the incoming preproteins, or at least associates with the TIC translocon and for a long time, this scenario could not be shown for stromal Hsp70, hence it seemed unlikely that Hsp70 alone provides the required power. However, it could be shown in 2010 for the moss P. patens that Hsp70 is indeed involved in protein import into chloroplasts as a stromal Hsp70 co-immunoprecipitated with early-import intermediates, as well as with Tic40 and Hsp93 (Shi and Theg, 2010). In agreement with this, Arabidopsis mutants lacking the chloroplast isoforms of Hsp70 showed a reduced import level of preproteins, which could also be demonstrated in

the moss Physcomitrella patens (Su and Li, 2010; Shi and Theg, 2010). Furthermore, it was suggested that the ATP requirements correlate with the activity of moss Hsp70, emphasizing the idea that cpHsp70 is the only energy-providing motor, at least in moss (Shi and Theg, 2010). Interestingly, Arabidopsis double mutants of Hsp93 and Hsp70 showed an additive effect in decreased import capacity compared to the single knockout mutants, leading to the theory that both proteins are acting at least partially in parallel as independent import players (Su and Li, 2010). This idea was somewhat supported later on: it was hypothesized that Hsp70 is the motor protein whereas Hsp93 is stably associated with the Clp protease complex at the IEM, suggesting a permanent role in quality control and degradation of preproteins and not a role in powering protein translocation (**Figure 3A**). In this study, the authors used a transgenic line in which the interaction of Hsp93 with the protease ClpP was disrupted, but the protein itself was still localized to the IEM and interaction with Tic110 was also ensured (Flores-Pérez et al., 2016). This enabled the study of the role of Hsp93 in protein import independent from its role in proteolysis. However, the truncated version could not complement the hsp93 import defective phenotype, thus excluding the possibility of Hsp93 being the main motor functioning in protein import (Flores-Pérez et al., 2016).

In remarkable contrast to the above-mentioned observations, a recent study on that topic could show that Hsp93 directly binds to both the N-terminal cTP and the mature part of incoming preproteins, thus clearly indicating a role in early import stages and challenging the above-mentioned theory (Huang et al., 2015). These authors favor the hypothesis that both chaperones could prefer different regions of the preprotein and thus provide different modes of translocation force, which would result in additive import defects in the double mutants. This would also hold true if Hsp93 was to be the primary motor for the cTP and Hsp70 for the mature region (**Figure 3B**). Preprotein processing takes place during binding to Hsp93 and thus, binding to the mature protein is also detected. In their model, Hsp70 is entirely responsible for interacting with the mature protein, acting in parallel and one defined step after the action of Hsp93 (**Figure 3B**).

Taken together, and taking the described discrepancies into account, it remains unclear why the chloroplast evolved such a complex and divergent chaperone system in comparison to other subcellular compartments such as mitochondria or the ER. However, as the cytosolic chaperones display distinct preprotein affinities, it is still reasonable to say that the stromal counterparts do the same, while keeping the opportunity to react efficiently toward different import conditions resulting from potential environmental stimuli.

### POTENTIAL INVOLVEMENT OF IMPORT REGULATION IN PLANT ACCLIMATION

Translocation efficiency of chloroplast proteins is highly dependent on post-translational modifications, enabling the plant to react quickly and efficiently toward external stimuli. The above-mentioned import steps can be influenced by various regulation mechanisms, including redox-mediated circuits of both cytosolic and stromal pathways, and phosphorylationdependent activities.

### Redox-Sensing at the Outer Envelope Membrane

Redox-mediated communication and regulation within cellular processes had already been present in the prokaryotic ancestor, thus leading to a range of reduction- and oxidation-driven regulation in the organelle. One of the best-studied mechanisms in the bacterial ancestor is the bacterial disulfide bond (Dsb) that ensures accurate folding of periplastic proteins (Guddat et al., 1998). The central component is DsbA, which contains a highly redox-active CXXC motif and can bind to its substrate during protein import. Similar to this, a thiol-dependent oxidation mechanism has been addressed to the thylakoid lumen (Gopalan et al., 2004).

Since even mitochondrial intermembrane-destined proteins are imported in an oxidation-driven reaction by the mitochondrial disulfide relay (Herrmann et al., 2009), this might also be the case for chloroplast import. However, this field has only recently gained more attention. Redox-mediated regulation could be observed at different stages of the import process. Both in the OEM as well as in the IEM, import-involved proteins exhibit redox-active properties.

Import activity is highly stimulated in vitro upon the addition of reducing agents like DTT or TCEP, and impaired by oxidizing substrates, suggesting a potential role of cysteines and disulfide bridges (Stengel et al., 2009). Interestingly, all TOC components contain several conserved cysteines. Seven are found in Toc75, present in both vascular and non-vascular plants. The POTRA domain contains four of them. In the case of cytosolic-facing POTRA domains, these cysteines could be involved in redoxmediated regulation. Toc159 displays five cysteines; two of them are conserved and are located within the GTPase domain. One of these two is also present in the GTPase domain of Toc34 and is fairly exposed (Sun et al., 2002), suggesting a suitable target for redox reactions. In the last identified TOC component, Toc64, ten cysteines could be found, of which six are conserved through vascular and non-vascular plants.

In their reduced state, Toc159, Toc34 and Toc75 are loosely attached, harboring different reduced thiols and thus, forming the so-called 'active' TOC complex, prepared for preprotein recognition and binding (**Figure 4A**). Upon oxidation, intraand intermolecular disulfide bridges are generated, commonly between Toc159, Toc34 and Toc75, resulting in a heteromeric TOC complex (Seedorf et al., 1995). Different hypotheses concerning the mode of action have been suggested. Oxidized on the one hand, this bulky complex could inhibit the import rate by simply blocking the channel and thus preventing the entrance of incoming proteins (**Figure 4B**). However, another mechanism suggests that not only does channel blocking occur, but also the preprotein-binding capacity of the receptor proteins is already altered as the cysteines are located within the preprotein binding, the GTPase, domain (Stengel et al., 2010). Up to now,

this model the main energy is consumed by Hsp70 and not by Hsp93 (Flores-Pérez et al., 2016). A recent study suggest that Hsp93 interacts subsequently with incoming preprotein at the N-terminal cTP, whereas Hsp70 binds to the mature parts of the protein (Huang et al., 2015). This enable the two chaperone systems to interact at least partially in parallel with the preproteins. After completing of the import by processing the cTP, proteins are folded with the help of various chaperones like Cpn60 and Hsp70 (B).

blocked TOC complex which inhibits import of precursor proteins either by blocking the channel or altering the binding capacity of the receptor toward the preproteins (B).

all these experiments have been carried out by adding, reducing or oxidizing agents in vitro, and so far, a discrete physiological role is still missing. However, it is still reasonable to assume that changing environmental conditions led to different redox states in the cytosol, due to the production of reactive oxygen species for example, and hence affecting the redox modulation of the translocation apparatus.

### Redox Sensing at the Inner Envelope Membrane

Regarding the fact that the TOC complex could be regulated in a thiol-dependent mechanism, it can be supposed that this regulation is also effective for the translocase of the IEM.

Indeed, a thiol-dependent interaction between Tic110 and Tic40 has been observed, but its in vivo role has to be clarified (Stahl et al., 1999). Tic110 itself has been found to contain one or two regulatory disulfide bridges (Balsera et al., 2009). These intramolecular bridges could have a critical influence on the structure and function of the central TIC component. Switches between reduction and oxidation of these disulfide bridges could either lead to an open or closed formation of Tic110, respectively, and thereby limit the amount of incoming preproteins (**Figure 5**). The stromal thioredoxin family has been demonstrated to operate on disulfide bonds of Tic110 (Balsera et al., 2009). The redox state of thioredoxins is directly linked to both photosynthetic activity and other redox-dependent mechanisms in the organelle, thus it might act as a transport signal that eventually reaches the import machinery to regulate the chloroplast import rate. The intermembrane space protein Tic22 contains a conserved cysteine (Glaser et al., 2012), which could be involved in intramolecular disulfide bridges leading to dimerization of Tic22. Furthermore, since Tic110 exposes one cysteine into the IMS, a possible disulfide bond between the soluble Tic22 and the pore protein Tic110 during preprotein is also a hypothesis. However, no redox-mediated modulation has been reported so far and this hypothesis has to be addressed experimentally.

A direct read-out for the stromal redox state is the ratio between NADPH and NADP+. These reduction equivalents

deliver electrons which are required for enzymatic activities of many biosynthetic pathways within the organelle. All the required enzymes for a subset of different pathways have to be imported at a specific rate depending on the actual need within the organelles. Therefore, protein import activity must be regulated according to these requirements, which could be mediated by the stromal redox state. Independent studies have shown that the stromal redox state influences the import activity of a subset of preproteins (Stengel et al., 2009; Zhang et al., 2016). Interestingly, several components, namely Tic62, Tic55 and Tic32, have been shown to associate dynamically with the core complex, leading to the assumption that these transient TIC components act in a regulatory mechanism in response to the stromal redox state (Stengel et al., 2008). One prominent candidate was Tic62, which showed a triple-localization pattern, shuttling from a membrane associated state at the IEM as well as the thylakoids to the stroma in response to changing NADP+/NADPH ratios, and thus being able to mediate signals from the photosynthetically active thylakoids to the import machinery (Stengel et al., 2008). Since one important function of Tic62 in vascular plants is the binding of the chloroplasttargeted ferredoxin-NADP(+) oxidoreductase (FNR) to these membranes via specific C-terminal motifs (Alte et al., 2010), its shuttling could significantly influence electron-transfer processes from this photosynthetic enzyme to different acceptor proteins which could display a signal transduction chain. As Tic62 possesses a NADPH binding site and acts as a dehydrogenase in vitro, it might also be involved in a direct electron transfer onto yet unknown acceptor proteins (Stengel et al., 2008). Furthermore, another binding partner of FNR, named Trol, has been characterized. This thylakoid-localized protein harbors a similar single C-terminal extension as is found repeatedly in Tic62 and was demonstrated to interact with FNR (Juric´ et al., 2009; Lintala et al., 2014). It could be shown that Trol also associates with the IEM, thus it might also participate in the signal transduction chain involving Tic62/FNR (Juric´ et al., 2009). Interestingly, the FNR binding C-terminal motif is exclusively found in vascular plants, leading to the assumption that this regulatory mechanism has evolved later in evolution. This might suggest that, for all other plants, the ecological pressure was not high enough to evolve a system that regulates their protein import activity in response to changing stromal redox conditions, which is in contrast to the old evolutionary regulation mechanism of thiol oxidation. A second protein possibly involved in redox regulation is Tic32, which is another member of a dehydrogenase family capable of transferring electrons. Like Tic62, the affinity toward the TIC complex is lower under reduced conditions. Interestingly, Tic32 is also subject to calmodulin/Ca2+ dependent regulation. It could be shown that calmodulin (CaM) directly binds to Tic32, which promotes import, and that specific inhibition of this interaction decreased import efficiency (Chigri et al., 2006). Thus, two very different modes of action can regulate the TIC translocon (**Figure 5**). The third member of the redox regulon is Tic55, a Rieske protein found in close proximity of Tic110. It is anchored to the IEM by two alpha helices and exposes its C-terminal region into the stroma. Recently, Tic55 was identified as a potential thioredoxin target by affinity chromatography on a Trx-column (Bartsch et al., 2008), which is supported by the presence of a CXXC motif. The molecular function of Tic55 is still unclear, but recently, a study was published in which a hydroxylation activity during leafsenescence-dependent chlorophyll breakdown was demonstrated for Tic55 (Hauenstein et al., 2016). This potential function of Tic55 connects chlorophyll metabolism to the chloroplast import demand and could function as a coordinator of the chloroplast homeostasis, similar to GUN1, which is a mediator of retrograde signaling. Under stress conditions, when chlorophyll is degraded, Tic55 could relay the required information which would eventually reach the nucleus in order to respond efficiently toward external stimuli. All the presented import regulation mechanisms are clearly involved in fine-tuning of the process rather than representing a molecular on/off switch, since single knockout mutants of the redox regulon components have, so far, no reported defects in protein import (Bölter et al., 2015).

### Phosphorylation of the TOC Complex

The number of import sites per chloroplast was estimated, leading to different results: counting the number of radioactive mature proteins inside the organelle led to an estimated number of 3,500 (Friedman and Keegstra, 1989), whereas the approach using immunogold labeling of ultrathin sections with antibodies against main import components resulted in a higher number of 35,000 import sites (Morin and Soll, 1997). The discrepancy between these numbers can be explained by the fact that the immunogold labeling informs us about the total number of import complexes in the envelope, while the radioactive experiment gives us a measure of the fraction of these complexes that are actively importing. The switch between activity and nonactivity of import complexes is likely to be modulated by the number of preproteins in need of being imported, amongst other signals.

It has been suggested that heterodimerization of the receptors, as well as their preprotein-binding capacity, is regulated by phosphorylation. PsToc34 and atToc33 are phosphorylated, whereas atToc34 is not, giving the opportunity to hypothesize that this represents specificity toward a different subset of

preproteins (Jelic et al., 2003). The phosphorylation might negatively affect GTP and preprotein binding of the respective receptor, and the whole TOC integrity is negatively influenced by phosphorylation in vitro (Oreb et al., 2007).

Signals triggering phosphorylation are, however, still not well defined. Data from transgenic Arabidopsis mutant lines showed that a phosphomimicking mutant of atToc33 is indeed affected in import capacity, whereas a non-phosphorylatable version of atToc33 exhibited a WT-like phenotype (Aronsson et al., 2006; Oreb et al., 2007). The latter observation in particular clearly indicates that phosphorylation mediated regulation is not a common or permanent regulation mechanism during plant development but rather an on/off switch in response to either a short period of a developmental change or to different, yet undefined, external stimuli. This could be the case, for example, upon cold or high light stress where the protein demand in the chloroplast is changed, or a specific subset of proteins is required. Under these conditions, fast post-translational modification machinery is required and phosphorylation of the TOC receptors might represent a relevant and efficient target for such an event. Regulation could occur in two ways. On the one hand, the overall import rate is affected (reduced, if phosphorylated) by downregulating the affinity to preproteins. On the other hand, phosphorylation could change the TOC complex stability, which would lead to the import of a distinct subset of client proteins.

Under this aspect, it would make sense if the responsible kinase was located in close proximity to avoid long shuttling pathways and to ensure specificity. In pea chloroplasts a 98-kDa, ATP-dependent outer membrane-attached kinase was identified as the responsible kinase (Sveshnikova et al., 2000). However, an Arabidopsis homolog is still missing, thus the identity of the responsible kinase remains elusive. Besides atToc33 (psToc34), the Toc159 family is also a target for phosphorylation, which is putatively mediated by a 70-kDa, ATPdependent kinase (Fulgosi and Soll, 2002). All members are

### REFERENCES


highly phosphorylated in their variable A-domain, consequently leading to a distinct phosphorylation pattern, ranging from many phosphorylation sites (atToc159) to few (all others) (Agne et al., 2010). As the A-domain between the members already displays a heterogeneous profile in sequence characteristics, the phosphorylation event could either be irrelevant or, contrarily, even enhance specificity toward preproteins.

Besides having a direct effect, phosphorylation could also act as part of a signaling cascade or promote indirectly another posttranslational mechanism, like ubiquitination. It has been shown that phosphorylation can indeed have a negative or positive effect on ubiquitination (Hunter, 2007). This would provide a link to a recently made observation. Ling and Jarvis (2016) identified an OEM E3 ubiquitin ligase (SP1), which upon abiotic stress marks TOC components for degradation. It must be clarified if phosphorylation enhances this effect, which would provide new insights into the regulation made by phosphorylation.

### DEDICATION

We dedicate this review article to the memory of Kentaro Inoue and his significant contributions to the chloroplast import field.

### FUNDING

JS and BB received funding from the DFG (SFB TR175 B5 and B6, respectively).

### AUTHOR CONTRIBUTIONS

All authors listed have made substantial, direct and intellectual contribution to the work, and approved it for publication



the chloroplast envelope. Plant Physiol. 170, 147–162. doi: 10.1104/pp.15. 01538


import machinery of plastids. Plant Cell 17, 1482–1496. doi: 10.1105/tpc.105. 030700


of Toc75 in chloroplast protein import. eLife 5:e12631. doi: 10.7554/eLife. 12631


nucleus-encoded chloroplast proteins. J. Cell Biol. 165, 323–334. doi: 10.1083/ jcb.200311074


import receptor Tom70. Cell 112, 41–50. doi: 10.1016/S0092-8674(02) 01250-3


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Sjuts, Soll and Bölter. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Post-translational Modifications in Regulation of Chloroplast Function: Recent Advances

Magda Grabsztunowicz, Minna M. Koskela and Paula Mulo\*

Molecular Plant Biology, Department of Biochemistry, University of Turku, Turku, Finland

Post-translational modifications (PTMs) of proteins enable fast modulation of protein function in response to metabolic and environmental changes. Phosphorylation is known to play a major role in regulating distribution of light energy between the Photosystems (PS) I and II (state transitions) and in PSII repair cycle. In addition, thioredoxin-mediated redox regulation of Calvin cycle enzymes has been shown to determine the efficiency of carbon assimilation. Besides these well characterized modifications, recent methodological progress has enabled identification of numerous other types of PTMs in various plant compartments, including chloroplasts. To date, at least N-terminal and Lys acetylation, Lys methylation, Tyr nitration and S-nitrosylation, glutathionylation, sumoylation and glycosylation of chloroplast proteins have been described. These modifications impact DNA replication, control transcriptional efficiency, regulate translational machinery and affect metabolic activities within the chloroplast. Moreover, light reactions of photosynthesis as well as carbon assimilation are regulated at multiple levels by a number of PTMs. It is likely that future studies will reveal new metabolic pathways to be regulated by PTMs as well as detailed molecular mechanisms of PTM-mediated regulation.

### Edited by:

Dario Leister, Ludwig-Maximilians-Universität München, Germany

#### Reviewed by:

Markus Schwarzländer, University of Bonn, Germany Marc M. Nowaczyk, Ruhr University Bochum, Germany

> \*Correspondence: Paula Mulo pmulo@utu.fi

#### Specialty section:

This article was submitted to Plant Cell Biology, a section of the journal Frontiers in Plant Science

Received: 02 January 2017 Accepted: 08 February 2017 Published: 23 February 2017

#### Citation:

Grabsztunowicz M, Koskela MM and Mulo P (2017) Post-translational Modifications in Regulation of Chloroplast Function: Recent Advances. Front. Plant Sci. 8:240. doi: 10.3389/fpls.2017.00240 Keywords: chloroplast, phosphorylation, photosynthesis, post-translational modification, redox regulation

## INTRODUCTION

Chloroplasts are sites of versatile metabolism. In addition to photosynthetic reactions, chloroplasts host a number of other processes, such as nitrogen and sulfur assimilation, amino acid and fatty acid biosynthesis as well as accumulation of pigments, photoreceptors, and hormones. Chloroplasts are surrounded by the envelope membrane, and the majority of nuclear-encoded chloroplast proteins are imported through the envelope into the plastid via the Toc/Tic machinery. The subchloroplastic destination of a specific protein is determined by the information buried within the primary amino acid sequence, either in the form of cleavable transit peptide or as an internal targeting signal. Due to their endosymbiotic origin, biosynthesis and function of chloroplasts is not only dependent on nuclear control, but also on the expression of approximately 120 plastome encoded genes, mostly involved in photosynthesis and plastid gene expression (Sugiura, 1992; Green, 2011). Obviously, coordination of gene expression between these compartments as well

**Abbreviations:** PS, photosystem; psa, genes encoding subunits of Photosystem II; psb, genes encoding subunits of Photosystem II; PTM, post-translational modification; RA, Rubisco activase; RB, RNA binding protein; rps, genes encoding ribosome subunits; trn, genes encoding chloroplast transferRNAs.

as integration of plastid metabolism with the rest of the cell are required to induce appropriate physiological responses to various environmental stimuli, thereby enabling successful growth and reproduction of the plants. This coordination takes place at many different levels, including the control of nuclear and plastid transcription, RNA processing and translation, protein translocation and assembly of protein complexes as well as functional adjustments of specific enzymes and/or pathways.

Recent interest and methodological progress on PTMs of non-histone proteins has revealed that also a great number of chloroplast proteins are post-translationally modified, which denotes for covalent processing of a mature protein. The most well-studied chloroplast protein modifications: (de)phosphorylation, conveyed by kinases and phosphatases, and oxidation-reduction (including disulfide-thiol exchange of Cys residues, regulated via thioredoxins) have been extensively reviewed (e.g., Buchanan and Balmer, 2005; Tikkanen and Aro, 2012; Michelet et al., 2013; Rochaix, 2013) and thus are not described in detail in the present article. Other PTM types, such as acetylation, methylation, glycosylation, nitration and nitrosylation, sumoylation, and glutathionylation have been identified in chloroplast proteins much later. As only limited information is available for these PTMs, it is currently not possible to conclude whether a given PTM is found in the chloroplasts of all plant and algal species, or whether it is specific for a certain group of organisms. In addition to the PTMs modifying a given amino acid, recent studies have shown that a number of chloroplast proteins are prone to N-terminal trimming resulting in different N-termini or N-terminal acetylation (Lehtimäki et al., 2015; Rowland et al., 2015). It is intriguing that both nuclear- and chloroplast-encoded proteins may be subjected to these modifications (Lehtimäki et al., 2015). In most cases the site of the PTM (cytosol or plastid) and/or the responsible enzymes have remained obscure. PTMs alter the physicochemical properties and thus the function of proteins in different ways depending on the modification and the molecular environment. The molecular structures of the different chloroplast PTM are presented in **Figure 1**. Here, we will draw together the current understanding of the PTMs regulating distinct metabolic processes in chloroplasts, and review the known physiological effects of these modifications.

### CHLOROPLAST MACHINERY FOR DNA REPLICATION AND GENE EXPRESSION

Organellar genomes are organized as nucleoids (also called as transcriptionally active chromosomes or TACs), DNA-protein complexes, which have been identified as the sites for both DNA replication and transcription (Melonek et al., 2016). Recently, proteomic analyses have suggested that also mRNA processing, splicing, editing, and ribosome assembly occur in association with the nucleoid, which supports the idea of co-transcriptional translation of plastid-encoded genes (Majeran et al., 2012). Although only few examples are thoroughly studied, PTMs of various types have been shown to regulate chloroplast genome replication and gene expression at multiple levels (**Figure 2**).

Even if a glycosylation machinery has been identified only in endoplasmic reticulum, some chloroplast proteins have been found to be glycosylated suggesting an existence of vesicular Toc/Tic independent chloroplast protein import route (Villarejo et al., 2005). One of the glycosylated proteins is the pea chloroplast protein p43, which associates with and activates the chloroplast DNA polymerase (Chen et al., 1996). Specifically, the N-terminal domain of p43 is highly O-arabinosylated (Gaikwad et al., 1999). Glycosylation of the protein is required for the induction of polymerization activity, although DNA binding is retained even if the protein is deglycosylated (Gaikwad et al., 1999, 2000). In addition to DNA replication, transcriptional activity of chloroplast genes is (partly) regulated by PTMs. Two different types of RNA polymerases, the plastid-encoded polymerase PEP, and the nuclear-encoded polymerase NEP, are responsible for the transcription of plastidencoded genes (Shiina et al., 2005). The core subunits of PEP polymerase are associated with nuclear-encoded sigma factors, which are regulated by (de)phosphorylation (Link, 2003; Shimizu et al., 2010). Ser phosphorylation of the sigma factors (SIG6 being the most well studied one) is at least partly conveyed by the plastid transcription kinase (PTK), which is a chloroplast Ser/Thr protein kinase (Baginsky et al., 1997; Baena-González et al., 2001; Ogrzewalla et al., 2002; Salinas et al., 2006; Schweer et al., 2010). The kinase itself is regulated via autophosphorylation and glutathione-dependent redox regulation (Baginsky et al., 1997, 1999; Ogrzewalla et al., 2002). Effect of sigma factor phosphorylation on transcription depends on the sigma factor and the transcribed gene in question: for instance phosphorylation of the Thr<sup>170</sup> in SIG1 inhibits transcription of the psaA gene (Shimizu et al., 2010), while phosphorylation of Ser94/<sup>95</sup> and/or Ser<sup>174</sup> in SIG6 enhances transcription of the atpB and trnK genes with no apparent effect on the transcription of the psbA gene (Schweer et al., 2010).

Processing of the chloroplast transcripts is also affected by phosphorylation and redox regulation of RNA binding proteins. For instance phosphorylation of endoribonuclease p54, which is responsible for the 3<sup>0</sup> processing of the plastid trnK and rps16, affects the RNA processing activity but not the cleavage specificity (Nickelsen and Link, 1993; Liere and Link, 1994). Additionally, the processing activity of p54 was modulated by glutathione (Liere and Link, 1994). Phosphorylation of 24 kDa (24RNP) and 28 kDa (28RNP) RNA-binding proteins, associated with a complex regulating the maturation of the 3<sup>0</sup> end of chloroplast transcripts (Hayes et al., 1996), has been shown to affect the affinity of the proteins to RNA. Specifically, phosphorylation of 24RNP increased its binding capacity to petD and psbA 3 <sup>0</sup> UTR (Loza-Tavera et al., 2006), whereas phosphorylation of 28RNP resulted in decreased affinity to RNA (Lisitsky and Schuster, 1995). Recently, it was shown that phosphorylation status of the 24RNP and 28RNP (and apparently other unidentified RNA binding proteins) mediates the interplay between the petD mRNA stability and processing (Vargas-Suarez et al., 2013).

The translational machinery of the chloroplast is composed of prokaryotic-type 70S ribosomes organized in small (Yamaguchi et al., 2000, 2003) and large (Yamaguchi and Subramanian, 2000) subunits. Chloroplast ribosomes contain rRNA and proteins,

the N-terminal amino acid of a protein). (B) Lys acetylation. (C) Lys mono-, di- and trimethylation (Me denotes methyl group). (D) Ser and Thr phosphorylation (P denotes phospho group). (E) Tyr nitration. (F) S-nitrosylation. (G) Disulfide formation (-SS- denotes disulfide bridge). (H) Glutathionylation of Cys (GSH denotes reduced and G oxidized glutathione). (I) Lys sumoylation. (J) N-glycosylation of Asn (-N-X denotes N-linked glycosyl group). Dash line indicates where structures have been cut off.

which are encoded both by the nuclear and chloroplast genomes (Carroll, 2013). Several ribosomal proteins in chloroplasts are targets of extensive PTMs, including formyl group or formyl methionine removal, N- and C-terminal processing, acetylation and monomethylation of N-terminal amino acids, trimethylation of Lys (Kamp et al., 1987; Schmidt et al., 1992; Yamaguchi and Subramanian, 2000; Yamaguchi et al., 2000; Alban et al., 2014) as well as phosphorylation (Guitton et al., 1984; Posno et al., 1984; Wagner et al., 2006). Recently, the enzyme responsible for the trimethylation of the internal Lys in Arabidopsis plastid ribosomal protein L11 (RPL11) has been identified as PrmAlike (Protein Arg methyltransferase-like) protein (Alban et al., 2014; Mazzoleni et al., 2015). Although depletion of Arabidopsis PRMA-like gene did not result in any phenotypic effects, mapping of the trimethylated Lys on the surface of the RPL11 protein allows hypothesizing that methylation might influence the stalk

region, which is responsible for the recruitment of initiation, elongation and release factors (Mazzoleni et al., 2015).

A special case in the chloroplast gene expression processes is the regulation of psbA gene expression, which has been under intense study for decades. The psbA gene encodes the lightsensitive PSII core subunit D1, which is constantly degraded and resynthesized in a light-responsive PSII repair cycle (Aro et al., 1993; Mulo et al., 2008). It has been shown that in chloroplasts of green algae and higher plants psbA gene expression is mainly controlled at post-transcriptional levels (Mulo et al., 2012). In Chlamydomonas reinhardtii, ADP-dependent phosphorylation of the cPDI (chloroplast protein disulfide isomerase or RB60) protein in darkness leads to release of the protein from the 5 <sup>0</sup> UTR of psbA mRNA and cessation of translation (Danon and Mayfield, 1994). Additionally, binding of RB47 to the psbA mRNA is controlled via redox regulation of disulfide groups in RB60 (Danon and Mayfield, 1994; Alergand et al., 2006). It has also been hypothesized that phosphorylation of the spinach 28RNP (in addition to participating in 3<sup>0</sup> UTR processing, see above) and ribosomal protein(s) might provide a light-dependent translation control mechanism for the chloroplast, especially during the repair cycle of PSII (Lisitsky and Schuster, 1995; Trebitsh et al., 2000; Yamaguchi and Subramanian, 2003).

## LIGHT REACTIONS OF PHOTOSYNTHESIS

Light reactions of photosynthesis, i.e., capture of light energy by the light harvesting complex (LHC) for the production of reducing power (NADPH) occur at the thylakoid membrane via the thylakoid-embedded pigment-protein complexes, namely PSII, Cyt b6f, and PSI. Concomitantly, protons are pumped into the thylakoid lumen, and ADP is photophosphorylated to ATP upon release of the generated proton gradient via the ATP synthase (**Figure 3**). NADPH and ATP, in turn, are used for numerous reactions, carbon assimilation being the major process. PSII functions as an oxygen-plastoquinone oxidoreductase, which is prone to light-induced photoinhibition (Aro et al., 1993; Tyystjärvi, 2013). The PSII core proteins D1 and D2 as well as the inner antenna protein CP43 and a minor PSII subunit PsbH are targets for light-dependent Thr phosphorylation (**Figure 3**) catalyzed mainly by the STN8 kinase (Bellafiore et al., 2005; Bonardi et al., 2005; Fristedt and Vener, 2011), while the PSII CORE PHOSPHATASE is responsible for the reverse reaction (i.e., dephosphorylation; Samol et al., 2012). PSII protein phosphorylation is involved in the folding of the thylakoid membrane, which affects the lateral migration of damaged D1 protein from grana stacks to stroma lamellae for degradation and resynthesis (Tikkanen et al., 2008; Fristedt et al., 2009). Another well-studied phosphorylation process is involved in the balancing electron transfer between PSII and PSI according to ambient environmental cues (i.e., light quality and quantity). Phosphorylation of the light harvesting proteins Lhcb1, Lhcb2 and Lhcb4 is catalyzed by the STN7 kinase (Depege et al., 2003; Bellafiore et al., 2005), and instead of PSII, the phosphorylated LHC trimers deliver excitation energy to PSI (so called state transitions) to adjust the absorption cross sections of the two PSs (Rochaix, 2014). Dephosphorylation of LHC by the PPH1/TAP38 (chloroplast protein phosphatase/thylakoid associated phosphatase of 38 kDa) protein phosphatase, in turn, results in redistribution of excitation energy toward PSII (Pribil et al., 2010; Shapiguzov et al., 2010). The STN7 kinase is activated by the binding of plastoquinol to the Q<sup>o</sup> site of Cyt b6f complex

3-phosphoglycerate; 1,3BPG, 1,3-bisphosphoglycerate; G3P, glyceraldehyde 3-phosphate; DHAP, dihydroxyacetone phosphate; FBP, fructose 1,6-bisphosphate; F6P, fructose 6-phosphate; X5P, xylulose 5-phosphate; R5P, ribose 5-phosphate; Ru5P, ribulose 5-phosphate; RuBP, ribulose 1,5-bisphosphate; S7P, sedoheptulose 7-phosphate; SBP, sedoheptulose 1,7-bisphosphate; E4P, erythrose 4-phosphate. The modified proteins are indicated with colors, while the

non-modified proteins are shown as transparent. See text for details.

(Vener et al., 1997; Lemeille et al., 2009) and inhibited by stromal reductants (Rintamäki et al., 2000).

In addition to phosphorylation, LHC proteins are prone to various other PTMs (**Figure 3**), such as N-terminal acetylation (Michel et al., 1991; Wu et al., 2011; Rowland et al., 2015), Lys-acetylation (Finkemeier et al., 2011; Wu et al., 2011), Tyrnitration (Galetskiy et al., 2011b) and sumoylation (Elrouby and Coupland, 2010; López-Torrejón et al., 2013). As acetylation neutralizes the positive charge either on the protein N-terminus or on Lys residue, it has numerous implications in biologic processes including determination of enzyme activity, protein stability and mediation of protein–protein interactions (Hwang et al., 2010; Bienvenut et al., 2011; Scott et al., 2011; Hoshiyasu et al., 2013). Accordingly, acetylation of the Lhcb1 and Lhcb2 proteins appear to be involved in the regulation of LHC attachment to the PSII complexes: the peripheral LHC antenna loosely bound to PSII showed higher level of Lys acetylation than the PSII-LHCII supercomplexes (Wu et al., 2011). In contrast to phosphorylation, acetylation status did not respond to changes in illumination (Wu et al., 2011). It is also worth noting that only the N-terminally trimmed form of Lhcb5 starting with Leu<sup>38</sup> (other forms starting with Phe<sup>39</sup> or Ser40) were reported to be Lys acetylated, indicating a cross-talk between N-terminal processing and acetylation of chloroplast proteins (Wu et al., 2011). Neither the chloroplast acetylation machinery (Dinh et al., 2015) nor the enzymes responsible for N-terminal processing (Rowland et al., 2015) have been thoroughly characterized yet. Also Tyr nitration of proteins representing PSII (including D1), Cyt b6f, PSI as well as LHC has been detected (Galetskiy et al., 2011a,b). Protein Tyr nitration is a marker of nitrosative stress, and it can irreversibly modify the conformation of proteins thus affecting the catalytic activity and susceptibility to proteolysis (Corpas et al., 2007). Indeed, changes in light conditions resulted in variation in nitration levels in different PSII-LHCII complexes, suggesting that nitration might be involved in photodamage, disassembly of complexes and subsequent degradation of proteins (Galetskiy et al., 2011a). It has also been found that LHC may be posttranslationally modified by sumoylation (Elrouby and Coupland, 2010; López-Torrejón et al., 2013), which refers to covalent binding of the small ubiquitin-like modifier (SUMO) protein (Miura et al., 2007). Sumoylation has been implicated in the regulation of protein localization, interactions and catalytic activity (Vierstra, 2012). Obviously, the exact effects of these PTMs on the function of LHC require further studies.

### CARBON ASSIMILATION AND STARCH METABOLISM

The photosynthetic carbon reduction cycle, i.e., the Calvin cycle, is a multistep pathway in which redox equivalents and chemical energy (NADPH and ATP) originating from the light reactions is utilized for the reduction of atmospheric carbon dioxide into organic compounds. Calvin cycle involves 11 stromal enzymes, which catalyze 13 distinct reactions. In the first step, inorganic CO<sup>2</sup> is fixed by ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco), producing 3-phosphoglycerate (3PG), which is first phosphorylated and then reduced into glyceraldehyde-3-phosphate (G3P). G3P then exits from the Calvin cycle and is further used for the synthesis of more complex sugars, including starch that is the most abundant storage polyglucan in nature (Tetlow and Emes, 2014). Several Calvin cycle enzymes have been reported to be activated in light upon reduction of specific disulfide bonds by thioredoxin (Pedersen et al., 1966; Jensen and Bassham, 1968; Buchanan and Wolosiuk, 1976; Wolosiuk and Buchanan, 1976; Buchanan, 1980). In addition to redox regulation, all steps of CO<sup>2</sup> fixation and starch metabolism are carefully controlled by multiple (PTM-dependent) mechanisms which balance the rate of starch synthesis with the availability of energy and carbon in different plant tissues and under various environmental conditions (**Figure 3**).

### Rubisco

In terrestrial plants and green alga, Rubisco exists as a holocomplex composed of eight nuclear-encoded small subunits (RBCS) and eight plastid-encoded large subunits (RBCL). Among the other enzymes involved in Calvin cycle Rubisco has been reported as a target of reversible phosphorylation in many plant species (**Figure 3**) (Reiland et al., 2009, 2011; Facette et al., 2013; Wang et al., 2014; Roitinger et al., 2015), RBCL being phosphorylated in response to light (Budde and Randall, 1990; Wang et al., 2014). The RBCL and RBCS subunits of Rubisco have been shown to contain multiple phosphorylation sites (Cao et al., 2011; Wang et al., 2014). Phosphorylation of the highly conserved RBCL residues Ser208, Thr246, Tyr<sup>239</sup> and Thr330, located in the close proximity to RuBP binding site, might affect the catalytic activity of the enzyme (Lohrig et al., 2009; Hodges et al., 2013). Indeed, dephosphorylation of RBCL has been shown to result in decreased activity of the enzyme (Chen et al., 2011), perhaps via affecting the interaction between Rubisco and RA (Guitton and Mache, 1987; Aggarwal et al., 1993; Hodges et al., 2013). Moreover, it has been suggested that dephosphorylation of RBCL and/or RBCS may lead to dissociation of Rubisco holocomplex (Guitton and Mache, 1987; Aggarwal et al., 1993; Hodges et al., 2013).

Rubisco has also been found as a target of both N-terminal acetylation and Lys acetylation (**Figure 3**). In spinach, RBCL is post-translationally processed by removal of Met<sup>1</sup> and Ser<sup>2</sup> followed by the acetylation of the penultimate amino acid (Mulligan et al., 1988). Although N-acetylation of proteins in general is known to modify their activity and stability, the detailed significance and mechanism of RBCL N-termini modification remains unknown (Mulligan et al., 1988; Houtz et al., 1992; Zybailov et al., 2008). Lys acetylation of the Rubisco subunits has been identified only recently, and it has been reported as a dynamic modification in response to the changes in the energy status in plants under different light conditions (Gao et al., 2016). The Rubisco holocomplex contains multiple Lys acetylation sites (e.g., nine in Arabidopsis and thirteen in wheat; Finkemeier et al., 2011), which are localized either in the catalytic center of Rubisco (Cleland et al., 1998; Finkemeier et al., 2011), at the interface between the two RBCL subunits (Knight et al., 1990; Finkemeier et al., 2011) or at the site crucial for

the formation of tertiary structure of Rubisco (Knight et al., 1990). Therefore, Lys-acetylation has been suspected to affect Rubisco activity and interactions between the subunits and with other molecules, and indeed recent studies have shown negative regulation of Rubisco activity by Lys acetylation (Finkemeier et al., 2011; Gao et al., 2016). Thus, acetylation of Rubisco might provide a mechanism to coordinate the function of light reactions and carbon assimilation with the carbon status of the cell.

In addition to acetylation, Lys residues of RBCL may be methylated (**Figure 3**). In many organisms (e.g., pea and tobacco), RBCL is considered as the main stromal methylprotein (Alban et al., 2014). Trimethylation of RBCL at Lys<sup>14</sup> has been found in several plant species (Alban et al., 2014; Ma et al., 2016) as a modification catalyzed by the large subunit Rubisco methyltransferase (LSMT), a highly conserved SETdomain protein lysine methyltransferase found in all plant species (Dirk et al., 2006). Despite numerous studies, the role of Lys<sup>14</sup> trimethylation of RBCL (as well as the role of methylation for other chloroplastic methylproteins) has not been identified (Clarke, 2013; Ma et al., 2016). Interestingly, in Arabidopsis, spinach, and wheat plants RBCL is not methylated at Lys<sup>14</sup> indicating species-specific differences in regulatory mechanisms (Houtz et al., 1992; Mininno et al., 2012; Ma et al., 2016). The methylation of chloroplast proteins seems to be biologically important, as a mutant impaired in PTAC14 (plastid-located SET-domain methyltransferase) exhibits defects in chloroplast differentiation and shows an albino phenotype (Steiner et al., 2011). On the other hand, the LSMT knockdown plants do not show any decrease in CO<sup>2</sup> assimilation and growth (Mininno et al., 2012).

Intriguingly, some Calvin cycle enzymes, including Rubisco, have been reported to be modified by peroxynitrite (**Figure 3**) (Cecconi et al., 2009; Lozano-Juste et al., 2011; Barroso et al., 2013). It has been suggested that Tyr-nitration of RBCL (and RA) might act as a modulator of plant defense-related responses including hypersensitive responses (Cecconi et al., 2009). On the other hand, Tyr-nitration of abundant proteins such as those involved in carbon metabolism might function as a nonspecific scavenging system for reactive nitrogen forms under stress conditions. Reversibility of Tyr-nitration is still discussed, thus additionally raising new questions about a potential function as a specific signaling event (Souza et al., 2008; Baudouin, 2011).

The reversible S-nitrosylation of Rubisco Cys residues has been reported both in vitro and in vivo for several plant species in response to nitric oxide (NO) -releasing compounds or to abiotic stresses (**Figure 3**) (Abat et al., 2008; Abat and Deswal, 2009; Fares et al., 2011; Vanzo et al., 2016). As the redox-active thiols in Cys residues can be modified by the covalent binding of NO resulting in the formation of S-nitrosothiol (Lindermayr et al., 2005), it is plausible that S-nitrosylation of Cys residues adjacent to the Rubisco active site in Arabidopsis might regulate the activity of the enzyme and degradation of the protein (Takahashi and Yamasaki, 2002; Marcus et al., 2003; Romero-Puertas et al., 2008). Indeed, recent enzymatic activity assays have revealed that Rubisco inactivation in response to S-nitrosylation is probably the main cause of reduction in carbon fixation upon various stress conditions (Clark et al., 2000; Abat et al., 2008; Abat and Deswal, 2009).

Another modification of Cys residues is protein S-glutathionylation, a well-described mechanism of signal transduction and protein regulation in mammals (Chrestensen et al., 2000). S-glutathionylation is a reversible post-translational formation of a mixed disulfide between the Cys residue of protein and glutathione. Previously, three Cys residues in RBCL and one in RBCS have been identified as targets of S-glutathionylation in plants (Rouhier et al., 2005), green alga (Zaffagnini et al., 2012a) and cyanobacteria (Sakr et al., 2013; Chardonnet et al., 2015). Protein S-glutathionylation probably protects specific Cys residues against irreversible oxidation under stress conditions (Ito et al., 2003; Zaffagnini et al., 2012b), but this PTM can also result in modulation of protein activity (Klatt and Lamas, 2000; Fratelli et al., 2004) and localization (Chardonnet et al., 2015). Nevertheless, the functional significance of Rubisco S-glutathionylation is not known yet.

### Activation and Function of the Calvin Cycle

Although PTMs of Rubisco have been extensively studied, also numerous other enzymes involved in carbon assimilation have been shown to possess multiple PTMs (**Figure 3**). As RA is responsible for removing inhibitors from Rubisco active center and thus contributes to initiation of carbon fixation, the stimuli affecting the RA activity is reflected in the yield of the entire carbon assimilation cycle. In green alga C. reinhardtii, RA is phosphorylated at Ser<sup>53</sup> by the thylakoid-localized Stn7 ortholog Stt7 kinase (see above; Lemeille et al., 2010). RA is mainly localized in the stroma, but a smaller portion of the enzyme has been found in association with the thylakoid membrane (Jin et al., 2006). It has been suggested that phosphorylation of RA increases the attachment of RA to the membrane, protecting Stt7 against proteolysis (Lemeille et al., 2009, 2010). The relocation could also be a mechanism reducing the activity of Rubisco under specific environmental conditions (Lemeille et al., 2010). In Arabidopsis plants, RA is phosphorylated at two sites, Thr<sup>78</sup> and Ser<sup>172</sup> (Boex-Fontvieille et al., 2014). In the dark, the phosphorylation percentage of Thr<sup>78</sup> increases (Reiland et al., 2009; Kim et al., 2016). As Thr<sup>78</sup> is located in the region crucial for Rubisco interaction (Zhang and Portis, 1999; Kim et al., 2016), it has been suggested that Thr<sup>78</sup> phosphorylation inhibits Rubisco activation (van de Loo and Salvucci, 1996; Stotz et al., 2011; Boex-Fontvieille et al., 2014). However, the importance of Thr<sup>78</sup> phosphorylation for the Rubisco activation requires further investigation as the Thr<sup>78</sup> is not conserved and replaced by Ile in maize and rice (Baginsky, 2016).

In addition to Rubisco, three other enzymes involved in Calvin cycle have been reported as phosphoproteins. Phosphoglycerate kinase (PGK) enzyme catalyzing the transfer of phosphate group from ATP to 3PG is phosphorylated in Arabidopsis, rice, and maize plants (Reiland et al., 2009; Facette et al., 2013; Roitinger et al., 2015; Baginsky, 2016). The two latter species

share the identical phosphorylation site VGAVSpSPK whereas in Arabidopsis PGK is phosphorylated in a domain much closer to the N-terminus. The kinase responsible for phosphorylation is unknown, but the phosphorylation motif suggests prolinedirected kinase as a possible candidate (Baginsky, 2016). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) possesses several phosphorylation sites, but as the sites differ significantly between different organisms, it is plausible that phosphorylation is not a major determinant of GAPDH activity in chloroplasts (Baginsky, 2016). Moreover, the main transketolase isoform in Arabidopsis (TKL1) is phosphorylated in a Ca2<sup>+</sup> dependent manner at Ser428, and phosphorylation affects enzyme activity (Rocha et al., 2014). Although Ser<sup>428</sup> is conserved in higher plants, Ser<sup>428</sup> has been found phosphorylated only in Arabidopsis plants (Hou et al., 2015; Baginsky, 2016). It is intriguing that PGK and GAPDH are also targets of Lys acetylation and S-glutathionylation (Finkemeier et al., 2011; Zaffagnini et al., 2012a; Chardonnet et al., 2015; Shen et al., 2015). The enzymatic activity of GAPDH and PGK is increased upon deacetylation, but the functional importance of S-glutathionylation of GAPDH and PGK remains to be elucidated (Finkemeier et al., 2011; Shen et al., 2015). These examples indicate that further studies are urgently needed in order to fully understand the dynamic regulation of Calvin cycle enzymes and to pinpoint the responsible enzymes involved (Friso and van Wijk, 2015; Baginsky, 2016).

Furthermore, a number of other enzymes involved in carbon assimilation have been shown to be post-translationally modified. For instance, fructose 1,6-bisphosphate aldolase (FBA) is trimethylated at a conserved Lys residue close to the C-terminus of the protein, however, without any effect on catalytic activity or the oligomeric state of the enzyme (Mininno et al., 2012; Ma et al., 2016). In poplar trees sedoheptulose-bisphosphatase (SBPase), RA, ribose-5-phosphate isomerase (RPI), phosphoribulokinase (PRK), GAPDH, triosephosphate isomerase (TPI), and PGK were S-nitrosylated during short-term oxidative stress induced by NO treatment (Vanzo et al., 2014, 2016), but the functional importance has not been described yet (Lindermayr et al., 2005; Abat et al., 2008; Romero-Puertas et al., 2008; Abat and Deswal, 2009).

### Starch Metabolism

Starch synthesis and degradation occur in a coordinated manner on a diurnal basis. In leaves, starch is synthesized during the day and degraded in darkness (Kötting et al., 2010). Reversible protein phosphorylation plays an important role also in the regulation of starch metabolism (Tetlow et al., 2004a, 2008; Grimaud et al., 2008; Reiland et al., 2009), and five different phosphoproteins (phosphoglucose isomerase, phosphoglucomutase, starch synthase and two subunits of ADPglucose pyrophosphorylase) involved in starch biosynthesis have been identified in Arabidopsis leaves (Geigenberger, 2011). Interestingly, starch synthase has been reported to be phosphorylated in a light dependent manner, i.e., exclusively at the end of the dark period (Reiland et al., 2009). Analyses of amyloplasts and chloroplasts from Triticum aestivum (wheat) have shown that some isoforms of starch-branching enzymes (SBE) are catalytically activated by phosphorylation and deactivated by dephosphorylation of one or more of their Ser residues (Tetlow et al., 2004b). Additionally, phosphorylation is apparently involved in the formation of protein complexes composed of starch synthase, SBE isoforms as well as other enzymes with undefined role(s) (Tetlow et al., 2004b; Kötting et al., 2010). It has been speculated that the physical association of the enzymes could alter their activities thus improving the efficiency of starch polymer construction (Kötting et al., 2010; Geigenberger, 2011). Moreover, numerous enzymes involved in starch metabolism, such as glucan water dikinase (GWD, also termed SEX1), starch excess4 (SEX4), β-amylase 1 (BAM1), ADP-glucose pyrophosphorylase, ADP-Glc transporter and class II SBE (Mikkelsen et al., 2005; Balmer et al., 2006; Sokolov et al., 2006; Valerio et al., 2011; Tuncel et al., 2014) are redox activated by thioredoxin. However, it is worth noting that redox modification of starch biosynthesis enzymes in response to light (and other environmental stimuli; reviewed in Kötting et al., 2010; Geigenberger, 2011) is not the only determinant of starch accumulation in plants, but most probably other (PTMdependent) regulatory mechanisms will be identified in the future (Li et al., 2012).

## CONCLUSION

Recently developed new experimental tools, i.e., PTM-specific antibodies and stains as well as enrichment techniques and high quality equipment for mass spectrometry have enabled identification of a range of PTMs in chloroplast proteins. Detailed knowledge about the effects of protein phosphorylation and redox regulation on the photosynthetic reactions already exists, but the regulation of most metabolic pathways in the chloroplast is poorly understood. Because a specific amino acid residue may be targeted by different PTM types (e.g., Lys methylation or Lys acetylation), and because different PTMs may have either antagonistic or cooperative effects, it will be important to reveal the entire PTM code of a protein(s) in order to understand the physiological significance of PTM-mediated regulation in a given metabolic pathway. Future studies are likely to reveal novel modification types as well as molecular mechanisms of PTM-dependent regulation of various metabolic pathways in chloroplasts.

### AUTHOR CONTRIBUTIONS

PM, MG, and MK have made substantial intellectual contribution to the work, participated in writing and revised the paper. MK and MG have drawn the figures. All authors have approved the paper for publication.

## FUNDING

This study was financially supported by Academy of Finland (307335 "Centre of Excellence in Molecular Biology of Primary Producers") and The Doctoral Programme in Molecular Life Sciences at the University of Turku.

### REFERENCES

fpls-08-00240 February 21, 2017 Time: 14:56 # 9



carboxylase/oxygenase activity and degradation. J. Bacteriol. 185, 1509–1517. doi: 10.1128/JB.185.5.1509-1517.2003


paradigm for the metabolic control of chloroplast carbon metabolism. Biochem. J. 458, 313–322. doi: 10.1042/BJ20130631



signal transduction between light and protein phosphorylation: thylakoid protein kinase deactivation by a single-turnover flash. Proc. Natl. Acad. Sci. U.S.A. 94, 1585–1590.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Grabsztunowicz, Koskela and Mulo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Maturation of Plastid c-type Cytochromes

#### Stéphane T. Gabilly1,2† and Patrice P. Hamel1,2 \*

<sup>1</sup> Department of Molecular Genetics and Department of Biological Chemistry and Pharmacology, The Ohio State University, Columbus, OH, United States, <sup>2</sup> Molecular and Cellular Developmental Biology Graduate Program, The Ohio State University, Columbus, OH, United States

Cytochromes c are hemoproteins, with the prosthetic group covalently linked to the apoprotein, which function as electron carriers. A class of cytochromes c is defined by a CXXCH heme-binding motif where the cysteines form thioether bonds with the vinyl groups of heme. Plastids are known to contain up to three cytochromes c. The membrane-bound cytochrome f and soluble cytochrome c<sup>6</sup> operate in photosynthesis while the activity of soluble cytochrome c6A remains unknown. Conversion of apoto holocytochrome c occurs in the thylakoid lumen and requires the independent transport of apocytochrome and heme across the thylakoid membrane followed by the stereospecific attachment of ferroheme via thioether linkages. Attachment of heme to apoforms of plastid cytochromes c is dependent upon the products of the CCS (for cytochrome c synthesis) genes, first uncovered via genetic analysis of photosynthetic deficient mutants in the green alga Chlamydomonas reinhardtii. The CCS pathway also occurs in cyanobacteria and several bacteria. CcsA and CCS1, the signature components of the CCS pathway are polytopic membrane proteins proposed to operate in the delivery of heme from the stroma to the lumen, and also in the catalysis of the heme ligation reaction. CCDA, CCS4, and CCS5 are components of trans-thylakoid pathways that deliver reducing equivalents in order to maintain the heme-binding cysteines in a reduced form prior to thioether bond formation. While only four CCS components are needed in bacteria, at least eight components are required for plastid cytochrome c assembly, suggesting the biochemistry of thioether formation is more nuanced in the plastid system.

### Edited by:

Dario Leister, Ludwig-Maximilians-Universität München, Germany

### Reviewed by:

Fevzi Daldal, University of Pennsylvania, United States Jean-David Rochaix, Université de Genève, Switzerland

> \*Correspondence: Patrice P. Hamel hamel.16@osu.edu

#### †Present address:

Stéphane T. Gabilly, Department of Plant and Microbial Biology, University of California, Berkeley, Berkeley, CA, United States

#### Specialty section:

This article was submitted to Plant Cell Biology, a section of the journal Frontiers in Plant Science

Received: 16 April 2017 Accepted: 12 July 2017 Published: 26 July 2017

#### Citation:

Gabilly ST and Hamel PP (2017) Maturation of Plastid c-type Cytochromes. Front. Plant Sci. 8:1313. doi: 10.3389/fpls.2017.01313 Keywords: thylakoid lumen, photosynthesis, cytochrome c, heme, thioether, cytochrome b6f

## CYTOCHROMES OF THE c-TYPE

Cytochromes of the c-type, often generically referred to as cytochromes c, are membranebound or soluble metalloproteins occurring in energy-transducing membranes in archaea, bacteria, mitochondria, and plastids, where they function as electron carriers in respiration and photosynthesis (Thony-Meyer, 1997; Bonnard et al., 2010; Kletzin et al., 2015). Cytochromes c, on the positive side (or p-side)<sup>1</sup> of energy-transducing membranes, carry one or several hemes (ferroprotoporphyrin IX) as a prosthetic group covalently attached via thioether bonds to a heme

<sup>1</sup>The p-side corresponds to the bacterial or archaeal periplasm, the thylakoid lumen and the mitochondria intermembrane space.

binding site in the apoprotein. The most common heme binding site consists of a CXXCH motif where the first and second cysteines are, respectively, linked to the vinyl-2 and vinyl-4 groups of heme and the intervening residue X can be any aminoacid except cysteine in naturally occurring cytochromes c (Allen et al., 2004; Bowman and Bren, 2008). The histidine residue serves as the proximal axial ligand of the iron atom. A distant histidine, methionine, or, less commonly, other residues in the apocytochrome provide distal axial ligation of the heme group (Bowman and Bren, 2008).

Variations of the heme binding site are rare and one example is the A/FXXCH motif of mitochondrial cytochromes c in Euglenozoa that bind the vinyl-4 group of heme via a single thioether bond (Priest and Hajduk, 1992; Fülöp et al., 2009). Other non-canonical heme binding sites occur in bacterial cytochromes c and contain three, four, or fifteen intervening residues between the cysteines instead of two (Herbaud et al., 2000; Aragão et al., 2003; Hartshorne et al., 2007) or a lysine instead of a histidine as the proximal heme ligand (Einsle et al., 1999). Another atypical cytochrome c is also the only known example of a cytochrome c on the negative side (n-side)<sup>2</sup> of the membrane. This is cytochrome b<sup>6</sup> of the b6f complex in plastids and cyanobacteria and cytochrome b of the bc complex in firmicutes, which contain a heme covalently attached via a single thioether bond (de Vitry, 2011). The heme binding cysteine faces the n-side of the membrane and is not found in a motif, unlike other p-side localized cytochromes c. Moreover, there are no amino-acid residues serving as proximal and distal ligands of heme, differentiating this cytochrome c from all other c-type cytochromes occurring on the p-side (de Vitry, 2011).

### CYTOCHROME c MATURATION SYSTEMS

All p-side localized holocytochromes c are assembled on the p-side of the membrane. This requires the apoform and the heme moiety, both of which are transported independently across at least one biological membrane. Conversion of apocytochrome to its holoform requires free sulfhydryls at the CXXCH motif, provision of heme under the reduced form (ferroheme), and stereospecific attachment of the prosthetic group via catalysis of the thioether bond linkage (Mavridou et al., 2013; Travaglini-Allocatelli, 2013). Extensive genetic and biochemical analyses in bacteria, plants, and fungi revealed that the operation of three distinct assembly pathways called Systems I, II, and III is required for cytochrome c maturation, depending on the location (Kranz et al., 2009; Mavridou et al., 2013; Travaglini-Allocatelli, 2013; Verissimo and Daldal, 2014; Babbitt et al., 2015). The diversity of maturation systems is surprising, considering the biochemical requirements for heme attachment to apocytochrome c are believed to be universal and thioether bond formation appears, a priori, a simple chemical reaction (Bowman and Bren, 2008). Each System can be recognized by prototypical assembly factors but the number of such assembly factors and their features differ considerably among the different Systems (**Table 1**). An additional layer of complexity is the apparent "mosaic" distribution of Systems I, II, and III among organisms and the different energy transducing membranes (Mavridou et al., 2013; Travaglini-Allocatelli, 2013). Several evolutionary scenarios accounting for the origin and distribution of the different maturation systems have been proposed but the complexity of cytochrome c maturation as a biochemical process still remains mysterious (Bertini et al., 2007; Allen et al., 2008; Giegé et al., 2008; Kranz et al., 2009; Allen, 2011).

## PLASTID c-TYPE CYTOCHROMES

Three c-type cytochromes, have been identified within the thylakoid lumen of various plastids: the membrane-bound cytochrome f and the soluble cytochromes c<sup>6</sup> and c6A. While cytochrome f and c<sup>6</sup> are known to function as electron carriers in photosynthesis, cytochrome c6A function remains enigmatic despite having been discovered 15 years ago (Howe et al., 2006). All plastid cytochromes c contain a single heme attached to a CXXCH motif on the apoprotein. Cytochrome f, a catalytic subunit of the cytochrome b6f complex, is universal in all photosynthetic eukaryotes (and cyanobacteria) and is essential for photosynthesis (Martinez et al., 1994). Cytochrome c<sup>6</sup> is found in cyanobacteria and the plastid of eukaryotic algae, where it is widely distributed among green, red and brown algal lineages (Sandmann et al., 1983; Kerfeld and Krogmann, 1998). Cytochrome c<sup>6</sup> is involved in the transfer of electrons from cytochrome f of the cytochrome b6f complex to Photosystem I (Merchant and Dreyfuss, 1998). In green algae and cyanobacteria, cytochrome c<sup>6</sup> acts as a substitute for plastocyanin in Cu-deficient conditions (Merchant and Bogorad, 1987a,b). Cytochrome c6A occurs in the thylakoid lumen of land plants and green algae but appears absent in red algae and diatoms (Wastl et al., 2004).

Cytochrome c6A was discovered in Arabidopsis as a protein interacting with the lumen-localized immunophilin FKBP13 in a yeast two-hybrid screen (Gupta et al., 2002a,b; Buchanan and Luan, 2005). It was initially postulated that cytochrome c6A acts as a substitute for plastocyanin (Gupta et al., 2002a), as in green algae and cyanobacteria where cytochrome c<sup>6</sup> can replace plastocyanin (Merchant and Bogorad, 1987a,b). However, loss of cytochrome c6A in Arabidopsis has no visible phenotype even under Cu deficient conditions (Gupta et al., 2002a). Moreover, an Arabidopsis plastocyanin-deficient mutant is unable to grow photoautotrophically even when cytochrome c6A is overexpressed (Weigel et al., 2003). In vitro, cytochrome c6A is unable to provide electrons to Photosystem I (Molina-Heredia et al., 2003). This observation accounts for the fact that cytochrome c6A cannot act as a functional substitute for plastocyanin in vivo. Hence, cytochrome c6A does not appear to function in the known electron transfer reactions of photosynthesis, which is consistent with its extremely low abundance in the thylakoid. The presence of a disulfide bond in holocytochrome c6A led to the proposal that the molecule acts as an oxidant of luminal proteins dithiols with heme providing

<sup>2</sup>The n-side corresponds to the bacterial or archaeal cytoplasm, the plastid stroma and the mitochondrial matrix.

#### TABLE 1 | Prototypical components of cytochrome c maturation pathways.


The prototypical components of Systems I, II, and III are indicated according to their proposed function in the maturation process. Bacterial cytochromes c are assembled via System I or II. Plastid cytochrome c assembly relies on System II and mitochondrial cytochromes c are matured via System I or III. System III is restricted to mitochondria and is defined by a single component, holocytochrome c synthase (HCCS). The nomenclature for Escherichia coli (System I), Bordetella pertussis (System II bacteria), and Chlamydomonas reinhardtii (System II plastids) is used here. With the exception of System II, for which a heme transport across the membrane is supported by experimental evidence, there is no description of transmembrane heme delivery routes in Systems I and III. '?' indicates that there is no component identified for this activity in the maturation process. System I: CcmABCDE<sup>1</sup> is a periplasmic heme handling route defined by an ABC transporter (CcmAB), a member of the Heme Handling Protein (HHP) family (CcmC), and a small transmembrane protein (CcmD) that are required to load heme onto a heme chaperone (CcmE) (Kranz et al., 2009). CcmF<sup>2</sup> and thiol-disulfide oxidoreductase CcmH<sup>3</sup> are forming a complex postulated to carry the holocytochrome c synthase activity (Kranz et al., 2009). System II: CcsB<sup>4</sup> is the bacterial ortholog of plastid CcsA.

electrons for re-oxidation of the cysteine pair (Marcaida et al., 2006). Additional experimental exploration is required to test this hypothesis.

### SYSTEM II, A MULTICOMPONENT ASSEMBLY PATHWAY REQUIRED FOR MATURATION OF PLASTID CYTOCHROMES c

Plastid cytochromes c are matured via System II, also referred to as the CCS pathway, a multicomponent assembly machinery (Hamel et al., 2009; Bonnard et al., 2010; Simon and Hederstedt, 2011). System II first emerged through genetic screens for photosynthesis-impaired ccs mutants (ccs for cytochrome c synthesis) in the green alga Chlamydomonas reinhardtii (Hamel et al., 2009; Simon and Hederstedt, 2011). The Chlamydomonas ccs mutants were isolated on the basis of a dual deficiency in the holoforms of both cytochrome f and cytochrome c6. All ccs mutants are photosynthetic deficient because loss of cytochrome f assembly results in a b6f-minus phenotype (Howe and Merchant, 1992; Inoue et al., 1997; Xie et al., 1998; Dreyfuss and Merchant, 1999; Page et al., 2004). Pulse-chase experiments revealed that both plastid apocytochromes c are synthesized, imported in the thylakoid lumen, and processed by lumenresident signal peptidase, but they fail to be converted to their respective holoforms (Howe and Merchant, 1993, 1994; Xie et al., 1998). Based on these experiments, it was concluded the ccs mutants exhibit a defect in the heme attachment to apoforms of cytochrome c in the thylakoid lumen (Howe and Merchant, 1993, 1994; Xie et al., 1998). The ccs mutants are also expected to display a defect in cytochrome c6A. However, this could not be tested because holocytochrome c6A could not be detected in a wild-type strain. The defect in the ccs mutants is specific to plastid c-type cytochromes since plastocyanin, another lumen-resident metalloprotein, is normally assembled (Howe and Merchant, 1993, 1994; Xie et al., 1998). The ccs mutants are not affected for the covalent attachment of heme to the single n-side facing cysteine in cytochrome b6, a structural subunit of the b6f complex (Kuras et al., 2007). Catalysis of this thioether bond in cytochrome b6, occurs on the stromal side of the thylakoid membrane and is dependent upon the CCB gene products (de Vitry, 2011).

### CCSA AND CCS1, A HEME DELIVERY COMPLEX WITH HOLOCYTOCHROME c SYNTHASE ACTIVITY

The first CCS component to be identified is plastid-encoded CcsA (Xie and Merchant, 1996; Hamel et al., 2003), a thylakoid membrane protein belonging to the HHP (Heme Handling Protein) superfamily (Lee et al., 2007), which is defined by the highly conserved tryptophan-rich WWD motif and conserved histidine residues (**Figure 1**). This feature is also shared by CcmC and CcmF, two HHPs in System I shown to relay heme on the bacterial periplasmic space (Richard-Fogal et al., 2009; Richard-Fogal and Kranz, 2010) (**Table 1**). The other prototypical component is CCS1, a thylakoid membrane protein with little sequence conservation and lacking domains or structural features speaking to a specific chemical function, with the exception of an invariant histidine (Inoue et al., 1997). Because all photosynthetic plastid genomes (with a few exceptions) encode a CcsA-like protein, the CCS pathway is believed to operate in the plastids of all photosynthetic eukaryotes. System II also occurs in cyanobacteria, a majority of the Gram-positive bacteria, proteobacteria of the β-, δ-, and ε groups, and aquificales (Hamel et al., 2009; Bonnard et al., 2010; Simon and Hederstedt, 2011).

Detailed studies, including topological studies of algal CcsA and cyanobacterial Ccs1, site-directed mutagenesis of conserved residues, and molecular analysis of existing ccs1 alleles, established that CcsA and CCS1 are polytopic membrane proteins with functional domains exposed to the lumen and four

is unclear, but it may be involved in the transport of reductant across the membrane. (2) A heme delivery/cytochrome c synthase pathway composed of CCS1 (dark blue) and CcsA (light blue). CcsA and CCS1 topologies were deduced using PhoA/LacZ topological reporters (Hamel et al., 2003; Dreyfuss et al., 2003). The CCS1/CcsA complex binds and channels the reduced heme from the stroma to the thylakoid lumen and catalyzes thioether bond formation. Two histidines, one from CCS1 and one from CcsA, on the stromal side of the thylakoid membrane could function in relaying the heme across the membrane. Two histidines in CcsA on the luminal side could act by coordinating heme that interacts with the WWD domain. Strictly conserved histidines (H), tryptophans (W), and asparatic acids (D) are circled and highlighted in yellow. The thylakoid membrane is drawn in gray.

strictly conserved essential histidine residues on both the lumen and stromal sides (Dreyfuss et al., 2003; Hamel et al., 2003) (**Figure 1**). Functional domains include the WWD signature motif for CcsA and a large hydrophilic C-terminal extension for CCS1. The C-terminal domain was postulated to chaperone apocytochrome c from studies of cyanobacterial Ccs1 (Tichy and Vermaas, 1999). In Chlamydomonas, a 200 kDa CCS1-containing complex in wild-type no longer accumulated in a ccsA mutant, suggesting that CcsA and CCS1 occur in a complex. The reduced abundance of CCS1 in some ccs mutants suggests the 200 kDa complex may contain other CCS components besides CcsA and CCS1 (Hamel et al., 2003). This led to the proposal that these two proteins act together to relay heme via histidinyl coordination from its site of synthesis, the stroma, to the lumen. In the lumen, heme is relayed to the WWD domain and coordinated by two histidine residues in CcsA (Hamel et al., 2003; **Figure 1**).

Experimental proof that CcsA and Ccs1 catalyze the heme attachment reaction was provided with the finding that Ccs1-CcsA fusion proteins, naturally occurring in several ε-proteobacteria, could assemble reporter cytochrome(s) c in an Escherichia coli strain lacking its endogenous cytochrome c assembly machinery (Feissner et al., 2006; Frawley and Kranz, 2009; Goddard et al., 2010; Kern et al., 2010; Richard-Fogal et al., 2012). Biochemical evidence supporting a possible role of CcsA and Ccs1 in heme transport from the cytoplasm to the periplasm came from studies of the Ccs1-CcsA fusion from Helicobacter hepaticus. Spectroscopic analysis of the recombinant fusion protein identified the presence of heme. Mutagenesis of the two periplasm-facing histidines highlighted the importance of these residues for the binding of heme and its maintenance in a reduced state (Frawley and Kranz, 2009). This led to the hypothesis that CcsA carries a heme binding site on the periplasmic space, presumably required for the cytochrome c synthase activity. To test the function of the two transmembrane cytoplasm-facing histidines in Ccs1 and CcsA, these residues were mutated in the recombinant protein. Because heme is synthesized in the cytoplasm and was no longer detected in the mutated form of the protein, it was concluded that these histidines provide an entry site for heme through the lipid bilayer on the cytoplasmic side of the membrane.

This implied that Ccs1-CcsA functions in channeling heme from the cytoplasm to the periplasm, but a direct heme transport activity remains to be demonstrated. By analogy, it is plausible that plastid CcsA and CCS1 also function in a heme relay pathway from stroma to lumen and carry the cytochrome c synthase activity but this has not been tested. While candidate components for the chemical reduction of heme were identified in System I, it is unknown how this process is achieved in System II (**Table 1**).

### OPERATION OF TRANS-THYLAKOID DISULFIDE REDUCING PATHWAYS

The operation of a thylakoid transmembrane thiol-disulfide relay in plastid cytochrome c maturation emerged with the description of two thiol-disulfide oxidoreductases at the

thylakoid membrane, namely CCDA, a member of the DsbD family, and HCF164, a membrane-anchored, lumen-facing protein that displays similarity to thioredoxin-like CcmG and CcsX (**Table 1**) (Lennartz et al., 2001; Page et al., 2004; Motohashi and Hisabori, 2006; Motohashi and Hisabori, 2010). In bacteria using Systems I and II, cytochrome c maturation requires the provision of reductants via sequential thiol-disulfide exchanges involving a cytoplasmic thioredoxin, a thiol-disulfide reductase of the DsbD family, and a periplasmic thioredoxin-like protein (CcmG in System I or CcsX in System II) (**Table 1**) (Mavridou et al., 2013; Travaglini-Allocatelli, 2013). The working model is that the apocytochrome c CXXCH motif is first disulfide bonded by the disulfide bond forming enzymes residing in the periplasm and subsequently reduced by a thioredoxin-like protein (CcmG or CcsX) dedicated to the heme attachment reaction (Mavridou et al., 2013; Travaglini-Allocatelli, 2013). Reverse-genetic analysis in Arabidopsis indicates a function for CCDA and HCF164 in holocytochrome f accumulation, but a possible defect in the heme attachment reaction was not investigated (Lennartz et al., 2001; Page et al., 2004).

The biochemical requirement for thiol-disulfide chemistry in plastid cytochrome c biogenesis was demonstrated with the identification of CCS5, the Chlamydomonas ortholog of thioredoxin-like HCF164 (Gabilly et al., 2010). CCS5 physically interacts with plastid apocytochromes c and a recombinant form of the CCS5 molecule is active as a reductase when apocytochrome c with a disulfide-bonded CXXCH motif is provided as a substrate in an in vitro reaction (Gabilly et al., 2010). Application of exogenous thiols to the ccs5-null mutant rescues the photosynthetic deficiency and holocytochrome f assembly, an indication that CCS5 acts as a disulfide reductase in vivo (Gabilly et al., 2010). By analogy to the bacterial pathway, CCS5/HCF164 is likely to be maintained reduced by the activity of CCDA but this remains to be experimentally tested (**Figure 1**). The source of reducing equivalents on the stromal side was attributed to thioredoxin-m (Trx-m) (**Figure 1**) based on the observation that the redox active cysteines in CCDA and HCF164 undergo reduction in isolated thylakoid membranes when Trx-m is added exogenously (Motohashi and Hisabori, 2010). Complete loss of function of CCDA or HCF164 in Arabidopsis and CCS5 in Chlamydomonas does not abolish plastid cytochrome c maturation, an indication that another mechanism for delivery of reductant must exist (Lennartz et al., 2001; Page et al., 2004; Gabilly et al., 2010).

Evidence of an additional pathway for the supply of reducing power was provided with the finding that the ccs4 mutant is restored for cytochrome c assembly by application of exogenous thiols (Gabilly et al., 2011). CCS4 is a small protein with an N-terminal membrane anchor and a C-terminal domain predicted to be exposed to the stromal side of the thylakoid membrane but does not display any motif or residue (such as cysteines) suggesting a role in thiol-based redox chemistry (Gabilly et al., 2011). CCS4 exhibits similarity to Arabidopsis HCF153, a thylakoid membrane anchored protein with a stromal facing C-terminal domain required for cytochrome b6f accumulation (Lennartz et al., 2006). In addition to the thiol-dependent photosynthetic rescue of the ccs4 mutant, the placement of CCS4 in a disulfide-reducing pathway for cytochrome c assembly is further substantiated by the fact that ectopic expression of CCDA, a thiol/disulfide oxidoreductase of the DsbD family, at the thylakoid membrane suppresses the ccs4 mutant (Gabilly et al., 2011). As none of the CCS loci correspond to CCDA (Page et al., 2004), the CCDA-dependent suppression of the ccs4 mutant provides indirect evidence for the function of CCDA in plastid cytochrome c maturation. The suppression can be explained by a compensatory effect due to enhanced expression of the thiol-disulfide oxidoreductase CCDA. The activity of CCS4 in the heme attachment reaction so far remains unclear but one attractive scenario is that it controls the delivery of reducing power through the membrane via transport of a reductant. There is precedence for this in bacterial periplasm where reducing power, in the form of cysteine or glutathione, is transferred from the cytoplasm to the periplasm via specific transporters (Pittman et al., 2005; Ohtsu et al., 2010).

### OTHER CCS COMPONENTS UNIQUE TO PLASTID CYTOCHROME C MATURATION

In bacteria using the CCS pathway, CcsA, Ccs1, a thiol-disulfide reductase of the DsbD family, and a thioredoxin-like protein are the only components required to complete holocytochrome c assembly (Beckett et al., 2000; Le Brun et al., 2000). From the genetic analysis of the Chlamydomonas ccs mutants, it appears that cytochrome c maturation in plastids is a more complicated process than in bacteria. This seems counterintuitive considering that bacteria can assemble numerous mono and multiheme cytochromes c via the CCS pathway, while plastids only need to mature up to three monoheme cytochromes. In addition to CcsA, CCS1, CCDA, and HCF164/CCS5, plastid cytochrome c maturation also requires CCS4 and the products of the CCS2, CCS3, and CCS6 genes (Xie et al., 1998; Page et al., 2004), which remain uncharacterized. The fact that single alleles map to the Chlamydomonas CCS3, CCS4, CCS5, and CCS6 loci suggests that mutant screens for plastid cytochrome c deficient mutants are not saturated and additional CCS loci could still be uncovered (Howe and Merchant, 1992; Xie et al., 1998; Dreyfuss and Merchant, 1999; Page et al., 2004).

### AUTHOR CONTRIBUTIONS

SG and PH wrote the manuscript jointly. SG designed the figure.

### ACKNOWLEDGMENTS

This work was primarily supported by a National Science Foundation grant (MCB-0920062) and a United States Department of Energy (DOE), Office of Science, Basic Energy Sciences (BES) grant (DE-SC0014562) to PH. We thank Dr. B. Alber, Dr. A. Simcox, Dr. S. Cline, M. Webber, and A. Das for critical reading of the manuscript.

### REFERENCES

fpls-08-01313 July 24, 2017 Time: 12:56 # 6



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Gabilly and Hamel. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# CCS2, an Octatricopeptide-Repeat Protein, Is Required for Plastid Cytochrome c Assembly in the Green Alga Chlamydomonas reinhardtii

Sara G. Cline1,2† , Isaac A. Laughbaum<sup>1</sup> and Patrice P. Hamel1,2 \*

<sup>1</sup> Department of Molecular Genetics and Department of Biological Chemistry and Pharmacology, The Ohio State University, Columbus, OH, United States, <sup>2</sup> Plant Cellular and Molecular Biology Graduate Program, The Ohio State University, Columbus, OH, United States

### Edited by:

Dario Leister, Ludwig-Maximilians-Universität München, Germany

#### Reviewed by:

Lixin Zhang, Institute of Botany (CAS), China Michael Schroda, Kaiserslautern University of Technology, Germany

> \*Correspondence: Patrice P. Hamel hamel.16@osu.edu

†Present address: Sara G. Cline,

Department of Mathematics, Computer and Natural Sciences, Athens State University, Athens, AL, United States

#### Specialty section:

This article was submitted to Plant Cell Biology, a section of the journal Frontiers in Plant Science

Received: 16 April 2017 Accepted: 12 July 2017 Published: 03 August 2017

#### Citation:

Cline SG, Laughbaum IA and Hamel PP (2017) CCS2, an Octatricopeptide-Repeat Protein, Is Required for Plastid Cytochrome c Assembly in the Green Alga Chlamydomonas reinhardtii. Front. Plant Sci. 8:1306. doi: 10.3389/fpls.2017.01306 In bacteria and energy generating organelles, c-type cytochromes are a class of universal electron carriers with a heme cofactor covalently linked via one or two thioether bonds to a heme binding site. The covalent attachment of heme to apocytochromes is a catalyzed process, taking place via three evolutionarily distinct assembly pathways (Systems I, II, III). System II was discovered in the green alga Chlamydomonas reinhardtii through the genetic analysis of the ccs mutants (cytochrome c synthesis), which display a block in the apo- to holo- form conversion of cytochrome f and c6, the thylakoid lumen resident c-type cytochromes functioning in photosynthesis. Here we show that the gene corresponding to the CCS2 locus encodes a 1,719 amino acid polypeptide and identify the molecular lesions in the ccs2-1 to ccs2-5 alleles. The CCS2 protein displays seven degenerate amino acid repeats, which are variations of the octatricopeptiderepeat motif (OPR) recently recognized in several nuclear-encoded proteins controlling the maturation, stability, or translation of chloroplast transcripts. A plastid site of action for CCS2 is inferred from the finding that GFP fused to the first 100 amino acids of the algal protein localizes to chloroplasts in Nicotiana benthamiana. We discuss the possible functions of CCS2 in the heme attachment reaction.

#### Keywords: plastid, photosynthesis, cytochrome c, heme, assembly factor, OPR

### INTRODUCTION

Energy-transducing membranes are specialized membranes in archaea, bacteria, mitochondria, and chloroplasts, which rely on electron carriers to generate the proton gradient necessary for ATP synthesis. In energy-transducing membranes, the c-type cytochromes, also generically referred to as cytochromes c, are a class of structurally diverse metalloproteins with one or more covalently linked heme(s) (ferro-protoporphyrin IX) that participate in electron transfer reactions (Thony-Meyer, 1997; Bonnard et al., 2010; Verissimo and Daldal, 2014). On the positive side (p-side) of the energy-transducing membrane (i.e., bacterial or archaeal periplasm, mitochondrial intermembrane space, thylakoid lumen), a group of cytochromes c occur as both soluble and membrane proteins with the heme moiety typically attached via a thioether linkage to two cysteine sulfhydryl(s) in a characteristic motif in the apocytochrome c. The consensus motif, also referred to as the heme-binding site, is CXnCH where n is usually equal to two and X can be any residue except

cysteine in naturally occurring cytochromes c. The histidine in this motif acts as a ligand of ferroheme (Bowman and Bren, 2008; Allen et al., 2009). Heme attachment to the heme-binding site is a catalyzed reaction in vivo and requires heme transport across at least one biological membrane and covalent linkage of ferroheme to free sulfhydryls of the heme binding cysteines in the CXnCH motif (Bonnard et al., 2010; Mavridou et al., 2013).

To date, three Systems (I, II, and III) have been identified as being necessary for the post-translational assembly of cytochromes c located on the p-side of the membrane, where these molecules function. These systems have been identified through extensive investigation of both bacterial and eukaryotic models (Kranz et al., 2009; Allen, 2011; de Vitry, 2011; Mavridou et al., 2013; Verissimo and Daldal, 2014). In plastids, the heme attachment reaction occurs on the luminal side of the thylakoid membrane and is under the control of System II. This system is composed of multiple pathways and occurs in plastids of all photosynthetic eukaryotes, cyanobacteria, and most proteobacteria of the β-, δ-, and ε-group (Hamel et al., 2009; Bonnard et al., 2010; Simon and Hederstedt, 2011).

System II first emerged through genetic screens for photosynthesis-impaired ccs mutants (ccs for **c**ytochrome **c s**ynthesis) in the green alga Chlamydomonas reinhardtii (Hamel et al., 2009; Simon and Hederstedt, 2011; Gabilly and Hamel, 2017). The ccs mutants were isolated on the basis of photosynthetic deficiency due to loss of plastid c-type cytochromes, namely membrane-bound cytochrome f and soluble cytochrome c<sup>6</sup> (Howe and Merchant, 1992). These two plastid cytochromes c reside in the thylakoid lumen and function in photosynthesis. The defect is specific to plastid cytochromes c, as abundance of mitochondrial cytochromes c was unaffected in the ccs mutants (Howe and Merchant, 1992). Pulse-chase analyses revealed that precursor forms of apocytochrome f and c<sup>6</sup> are synthesized, translocated to the lumen, and therein cleaved by the thylakoid peptidase. However, they are not converted to their respective holoforms. This indicates that the CCS loci control the heme attachment reaction in the lumen (Howe and Merchant, 1992, 1993, 1994; Xie et al., 1998; Gabilly et al., 2010). At least seven loci, plastid ccsA and nuclear CCS1 to CCS6, were uncovered through genetic analysis of the ccs mutants isolated via several UV and insertional mutageneses (Xie et al., 1998; Page et al., 2004; Gabilly et al., 2010).

Functional analysis of plastid CcsA and Ccs1 led to the proposal that these two proteins act together to relay heme from its site of synthesis, the stroma, to its site of function, the lumen, possibly in an assembly complex alongside other CCS factors (Xie and Merchant, 1996; Inoue et al., 1997; Dreyfuss et al., 2003; Hamel et al., 2003). Biochemical studies supported this model with evidence that bacterial CcsA and Ccs1 operate as a functional unit with a dual activity in heme transport across the membrane and heme ligation to apocytochromes c in the periplasmic space (Frawley and Kranz, 2009).

In addition to heme delivery and attachment, the maintenance of the cysteine sulfhydryls (reduced vs. oxidized) within the heme binding site is a prerequisite for covalent linkage of heme (Bonnard et al., 2010; Sanders et al., 2010; Verissimo and Daldal, 2014). The proposed model is that the cysteines are first oxidized by disulfide bond forming enzymes on the p-side of the membrane and must subsequently be reduced to provide free sulfhydryls for the heme ligation reaction. In plastids and bacteria, this requirement is fulfilled by the operation of a disulfide reducing pathway, which conveys electrons across the energy-transducing membrane via thiol-disulfide exchanges in order to reduce the disulfide bond formed between the hemelinking cysteines in the heme binding motif of apocytochromes c (Karamoko et al., 2013; Verissimo and Daldal, 2014; Gabilly and Hamel, 2017).

In bacteria, the CcsA/Ccs1-dependent heme delivery/attachment pathway and the disulfide-reducing pathway are all that is required for cytochrome c assembly (Beckett et al., 2000; Le Brun et al., 2000; Feissner et al., 2005). However, screening ofccs mutants in Chlamydomonas reinhardtii has revealed four additional CCS loci (CCS2, CCS3, CCS4, and CCS6), an indication that the assembly of cytochromes c in the plastid is a more complicated process (Xie et al., 1998; Page et al., 2004). While we have proposed that CCS4 regulates the disulfide reducing pathway in the plastid (Gabilly et al., 2011), the gene products for CCS2, CCS3, and CCS6 remain unknown.

In this article, we report the molecular identification of the CCS2 locus by functional complementation of the ccs2 mutant. The CCS2 gene encodes a plastid-localized, 170 kDa protein that was previously identified by bioinformatics to be a member of the **o**ctatrico**p**eptide **r**epeat (OPR) family (Eberhard et al., 2011).

### MATERIALS AND METHODS

### Strains and Culture Condition

The ccs2-1 through ccs2-5 strains were used in complementation experiments (Xie et al., 1998). Mutants ccs2-1 and ccs2-2 were crossed to wild-type strain 3A<sup>+</sup> (mt<sup>+</sup> arg7-8) and 4C<sup>−</sup> (mt<sup>−</sup> arg7-8), respectively, to generate the ccs2-1 arg7-8 and ccs2-2 arg7-8 recipient strains. The ccs2-1 arg7-8 mutant was then crossed to wild-type strain CC425 (arg7-8 cw15) to generate the ccs2-1 arg7-8 cw15 mutant used for detection of the HA-tagged CCS2 protein. The MCA1-HA expressing strain is described in Raynaud et al. (2007). Strains were maintained at 25◦C on Tris-acetate phosphate (TAP) liquid or solid medium supplemented with 400 mg/mL arginine (Harris, 1989) at 0.6 µE/m<sup>2</sup> /s. Complemented ccs2 strains were assessed for restoration of photoautotrophic growth on minimal medium (Min) (Harris, 1989) or acetate containing (1.7 mM) minimal medium. For protein extraction, wild-type and complemented strains were grown in liquid TAP supplemented with arginine under 50 µE/m<sup>2</sup> /s illumination while ccs mutants were grown under 0.6 µE/m<sup>2</sup> /s illumination. Cell wall-less mutants (cw15) were cultured in liquid and on solid media supplemented with 50 mM sorbitol. Copper-free media for induction of cytochrome c<sup>6</sup> was prepared as described previously (Howe and Merchant, 1992; Quinn and Merchant, 1998).

### Molecular Cloning of the CCS2 Gene

An indexed cosmid library of Chlamydomonas genomic DNA was used for transformation by electroporation as described

by Shimogawara et al. (1998) with the following exceptions: the vector backbone was the cosmid pCB412 containing the ARG7 marker and the transformation required a 30 min incubation in autolysin (to digest the cell wall) followed by electroporation of 5 µg of DNA per cosmid pool on a Biorad Micropulser at 1300 V. Transformants were selected under 50 µE/m<sup>2</sup> /s light on minimal media supplemented with 1.7 mM acetate. Plasmid pMOL+8.2 kb was generated by cloning the 8.2 kb BamHI/HindIII CCS2 genomic fragment isolated from complementing cosmid (c8G6), into BamHI/HindIII digested pMOLUC (Cha et al., 2002).

### Assembly of the CCS2 cDNA

Chlamydomonas RNAs were extracted and retro-transcribed using a bacterial reverse transcriptase from the Roche Transcriptor High Fidelity cDNA Synthesis Kit and the CCS2 specific primers CCS2.30 and CCS2.66STP. Overlapping fragments were amplified with the following primer pairs: CCS2.69 and CCS2.66STP, CCS2.19 and CCS2.54, CCS2.21 and CCS2.18, CCS2.79 and CCS2.02, CCS2.27 and CCS2.28, CCS2.51 and CCS2.26, and, finally, CCS2.81ATG and CCS2.70 using DV Ready Mix (Sigma) and the total retro-transcribed RNAs as a template. Amplified fragments were isolated after electrophoresis in agarose gel, purified, re-amplified and AT-cloned into pGEM-T Easy for sequencing. All primer sequences are listed in Supplementary Table S1.

### Construction of HA-Tagged CCS2 and CCS2-GFP Expressing Constructs

Versions of the CCS2 gene carrying an internal HA tag were created by cutting pMOL+8.2 kb by BsiWI or BspEI and inserting the HA-tag sequence via In-Fusion <sup>R</sup> (CloneTech). The HA-tag sequence was generated using PCR based fill-in of primers CCS2-HA\_BsiW1-F and CCS2-HA\_BsiW1-R or CCS2-HA\_BspEI-F and CCS2-HA\_BspEI-R. Introduction of an internal HA tag sequence at the BspEI site (residue 298) abolished CCS2 function while the tag at BsiWI (residue 1672) had no impact (not shown).

To generate the series of "p8" plasmids used in **Figure 5**, a 606 bp sequence was synthesized (Genscript) and cloned into pUC57 (pUC57+ccs2bit). This sequence contained two distinct fragments that were modified from the Chlamydomonas genomic DNA. The first was a 207 bp fragment corresponding to the 3<sup>0</sup> end of CCS2 genomic DNA that both changed the stop codon in the CCS2 ORF into an alanine and added the restriction sites XbaI, SwaI, and SpeI downstream of the stop codon. The second was a 394 bp fragment encompassing the 5<sup>0</sup> end of CCS2 genomic DNA designed to remove the native CCS2 promoter and introduce restriction sites BglII and XhoI upstream of the ATG.

Carboxy-terminus 3xHA tags were generated by cutting pMOL+8.2 kb by BsiWI (146 bp upstream the stop codon) and PshA1 (570 bp downstream of the stop codon in the 3<sup>0</sup> UTR) and cloning the 207 bp fragment, which was amplified by primers CCS2-BsiWI-F and CCS2-PshAI-R using pUC57+ccs2bits as a template. The 207 bp fragment was cloned using In-Fusion <sup>R</sup> and the resulting plasmid is p8-noS. The p8-noS plasmid was then digested by XbaI and SpeI and a sequence corresponding to a 3xHA tag (and including a stop), created through PCR fill-in of primers 8.3xHA-f and 8.3xHA-r, was inserted via In-Fusion <sup>R</sup> . This created vector p8-3xHA expressing CCS2-HA from its native promotor. The plasmids expressing CCS2 or CCS2-HA under the PSAD promoter were generated in a similar manner. The 394 bp fragment was amplified from pUC57+ccs2bits using primers CCS2-XcmI.R and CCS2-XcmI.F and inserted via In-Fusion <sup>R</sup> after digestion of pMOL+8.2 kb and p8-3xHA by XcmI (322 bp from the initiation codon). This created plasmids p8-noP and p8-3xHAnoP. These constructs were then used to generate plasmids p8-PROM and p8-3xHA+P. To create these vectors, the PSAD promoter and 5<sup>0</sup> UTR were amplified from plasmid pSL18 (Pollock et al., 2004), using primers 8.PROM-f and 8.PROM-r, and inserted via In-Fusion <sup>R</sup> at the introduced restriction sites BglII and XhoI.

The plasmid pGWB5/ccs2target, expressing the CCS2-GFP fusion protein, was constructed from pUC57+ccs2targeting, which contains the first 300 bp from CCS2 ORF, corresponding to the first 100 amino acids of CCS2, cloned into pUC57. This 300 bp sequence was codon optimized for expression in tobacco and synthesized by GenScript. Using primers CCS2.t1 and CCS2.t2, the sequence was amplified from the pUC57+ccs2targeting template and then inserted in frame with the GFP reporter in the expression vector pGWB5 (Nakagawa et al., 2007) using entry vector pENTR/SD/TOPO (Invitrogen) via TOPO cloning. The GFP-NLS/GFP-NES expressing construct in pK7WGF2 (Karimi et al., 2002) is a gift from Dr. I. Meier (Ohio State University). The NLS (nuclear localization sequence) is from the SV40 large T-antigen. The NES (nuclear export signal) is from the HIV-1 Rev response element. Plasmids were transferred to Agrobacterium tumefaciens GV3101 by electroporation.

### Protein Preparation and Analysis

Cytochrome f and c<sup>6</sup> detection was performed as in Howe and Merchant (1992). Soluble fractions for cytochrome c<sup>6</sup> detection were obtained by freeze-thaw fractionation of cells grown in copper deficient conditions. Fractions were electrophoretically separated and cytochromes c revealed by immunodetection or a heme-staining procedure (Howe and Merchant, 1992). Anti-HA immunoblotting analysis was performed on whole cells extracts, prepared as follows: cells were grown to early logarithmic phase under 0.6 µE/m<sup>2</sup> /s illumination and moved into 30 µE/m<sup>2</sup> /s light for 5 h. Cells were then pelleted and re-suspended in 10 mM NaPO<sup>4</sup> buffer with protease inhibitors (4 mM benzamine, 0.4 mM 6-amino-n-hexanoic acid, 2 mM PMSF, 10 µM leupeptin, 1 µM pepstatin, 1 mM ortho-phenanthroline, 40 µg/mL chymostatin, 10 µM E-64) (Kurvari et al., 1998) to a final concentration of 3 × 10<sup>8</sup> cells. Laemmli Buffer was then added to a final concentration of 1.2 × 10<sup>8</sup> cells and 0.1 M DTT. Cells were placed in a sonication bath for 120 s and the solubilized protein sample was denatured for 20 min at 70◦C before electrophoresis. Polyclonal antisera raised against Chlamydomonas cytochrome c6, cytochrome f GST-fusion protein, and CF<sup>1</sup> were used for immunodetection by alkaline phosphatase-conjugated secondary antibodies. The rat

Cline et al. Plastid Metalloprotein Assembly

monoclonal anti-HA antibody (clone 3F10) (Roche) was used for immunodetection by peroxidase-conjugated secondary antibody.

### Fluorescence Rise and Decay Kinetics

Fluorescence rise and decay kinetics were taken as described in Gabilly et al. (2010) except that strains were grown in liquid medium overnight and measurements were recorded on 20 µL culture aliquots against a white background. Fluorescence transients were measured using Handy Fluorcam from Photon System Instruments. The fluorescence is in arbitrary units (A.U.) and recorded over a 3-s illumination period.

### GFP Fluorescence and Imaging

Nicotiana benthamiana was transformed with Agrobacterium carrying the GFP-CCS2 or NLS-GFP/NES-GFP expressing construct via infiltration after a 1 h incubation in Induction Medium (10 mM MgCl2, 10 mM 2-[N-Morpholino]ethanesulfonic acid, 100 µM Acetosyringone). After 3 days, protoplasts were extracted from infiltrated leaves by a 30-min incubation in a Digestion Buffer (1.5% cellulose, 0.4% macerozyme, 0.4 M mannitol, 20 mM KCl, 20 mM 2-[N-Morpholino]ethanesulfonic acid -KOH pH 5.5, 10 mM CaCl2, 0.1% Bovine serum albumin) and concentrated with a 1 s spin at 100 rpm. Supernatant was removed and protoplasts were re-suspended in 100 µL Digestion Buffer before imaging. GFP-dependent fluorescence was taken at 515 nm and chlorophyll auto-fluorescence at 650 nm on a Nikon Cl confocal microscope (Eclipse C90i) using a medium aperture. Images were processed using the NIS-Elements software.

### Analysis of OPR Motifs in Proteins

Information presented in **Figure 4** was generated by the MEME algorithm (Bailey and Elkan, 1994). Individual protein sequences (Supplementary Figure S1) were entered into the program and base settings were altered to search for any number of repetitions, one motif, between 70 and 600 sites, and between 28 and 50 amino acids wide. The OPR motifs identified by the MEME program (Supplementary Figure S2) were fed into Weblogo 3.3 (Crooks et al., 2004).

### RESULTS

### Cloning of the CCS2 Gene by Functional Complementation of the ccs2-2 Mutant

To gain further insights into plastid cytochrome c assembly, we sought to clone the gene corresponding to the CCS2 locus, defined by the ccs2-1 to -5 alleles (Xie et al., 1998). All ccs mutants display a b6f-deficient phenotype due to loss of cytochrome f assembly and hence are unable to grow photoautotrophically (Howe and Merchant, 1992; Xie et al., 1998). Using the photosynthetic deficient phenotype of the ccs mutants, we cloned the CCS2 gene via transformation of the ccs2-2 arg7-8 double mutant with an indexed cosmid library (Purton and Rochaix, 1994; Zhang et al., 1994). Transformants were selected for restored growth on acetate-containing minimal medium

FIGURE 1 | The CCS2 gene restores photoautotropic growth to the ccs2 mutants. For (A,B), wild-type CC124 (WT), ccs2-1 and ccs2-2 strains, and CCS2 complemented mutants were used. (A) Ten-fold dilution series of algal cultures were plated on solid medium, with or without acetate, and incubated under low light (0.6 µE/m<sup>2</sup> /s) or high light (50 µE/m<sup>2</sup> /s), respectively, for 1 week at 25◦C. The ccs2-3, ccs2-4, and ccs2-5 mutants are also restored for photoautotrophic growth upon transformation with the CCS2 gene (not shown). (B) Representative fluorescence rise and decay kinetics indicate restoration of the cytochrome b6f complex in the complemented ccs2 mutant. While all complemented mutants were restored, only the ccs2-2 strain is shown. The fluorescence is in arbitrary units (A.U.) and recorded over a 3-s illumination period after a dark adaption period.

under standard illumination (50 µE/m<sup>2</sup> /s). Two cosmids (c8G6 and c5D9) were isolated based on their ability to restore photosynthetic growth to the ccs2-2 mutant. Sequence analysis revealed that c8G6 contains a 30.2 kb region from chromosome 19 while c5D9 appears to have rearranged. The complementing activity in c8G6 could be narrowed down to an 8.2 kb BamHI-HindIII fragment. Both c8G6 and the cloned 8.2 kb BamHI-HindIII fragment complemented strains ccs2-1 through ccs2-5, restoring photosynthetic growth (**Figure 1A**).

To test if the cytochrome f assembly was restored in the complemented transformants, we performed analyses of fluorescence rise and decay kinetics (**Figure 1B**). In such experiments, the emitted fluorescence of excited chlorophyll in photosystem II is taken as an indication of the functionality of the b6f complex, which receives electrons from photosystem II. As seen in **Figure 1B**, the rise and plateau curve for ccs2 is characteristic of a specific block in electron transfer at the level of the cytochrome b6f complex because of its impaired assembly in the absence of membrane-bound holocytochrome

FIGURE 2 | Restoration of plastid cytochrome c assembly in the CCS2 complemented strains. For (A,B), the ccs2-1 to ccs2-5 mutants were complemented with the 8.2 kb genomic fragment containing the CCS2 gene. Dilutions of the wild-type sample serve to estimate the cytochrome f (A) and cytochrome c<sup>6</sup> (B) abundance. (A) The wild-type CC124 strain (WT), ccs2 mutants, and two independently complemented ccs2 transformants for each ccs2 allele (CCS2) were analyzed for cytochrome f accumulation via heme stain and immunoblotting. Detection of the CF<sup>1</sup> of the ATPase is shown as a loading control. Note that heme staining also reveals the presence of covalently attached heme c<sup>i</sup> in holocytochrome b6. As evidenced from the heme stain, holocytochrome b<sup>6</sup> accumulates to a lower level due to loss of cytochrome f assembly in the ccs2 mutants. Different levels of holocytochrome f accumulation in the transformants might reflect differential expression of the introduced CCS2 gene due to position effect of non-homologous integration of the construct in the chromosome. (B) Same as in (A), except only one transformant was analyzed for cytochrome c<sup>6</sup> accumulation by heme staining and immunoblotting. Ponceau staining is shown as a loading control. The white lines in (A,B) indicate assembly from independent immunoblots.

f. When the energy absorbed by the chlorophyll cannot be utilized, in this case as a result of a block in photosynthetic transfer through cytochrome b6f, an increase in the chlorophyll fluorescence is observed. In wild-type and complemented strains, the decay phase corresponds to the re-oxidation of the quinone pool, the primary electron acceptor of the photosystem II, by the cytochrome b6f complex. This indicates holocytochrome f assembly and cytochrome b6f functionality is restored in the complemented strains. In accord with this result, we also showed that both holocytochrome f and cytochrome c<sup>6</sup> accumulation is restored to wild-type levels in the complemented strains (**Figures 2A,B**).

### The CCS2 Gene Encodes a Protein of the OPR Family

Because of the large size, low abundance, and high GC content (74%) of the CCS2 mRNA, the corresponding full-length cDNA proved difficult to amplify. Instead, overlapping cDNA fragments, approximately 2 kb in length, were amplified and the full-length transcript was extrapolated by aligning the sequenced fragments with the CCS2 genomic DNA. Sequence comparison of the assembled cDNA with the current gene model Cre03.g213201 extended the 5<sup>0</sup> end of exon 1 and identified four introns within the CCS2 gene (**Figure 3A**). The Chlamydomonas transcriptome from the Joint Genome Institute, University of California (JGI-UCLA) and Genoscope<sup>1</sup> corroborates this transcript sequence under the previous unique gene ID Cre19.g757200. We assigned the start codon in exon 1 based on the fact that this is the 5<sup>0</sup> most ATG preceded by stop codons in all three reading frames.

Sequencing of the genomic DNA in the ccs2-1 to ccs2-5 mutants revealed molecular lesions in the CCS2 coding sequence (**Figure 3A**). All lesions introduced nonsense (ccs2-1 and ccs2-3)

<sup>1</sup>http://genomes.mcdb.ucla.edu/Cre454/


FIGURE 3 | Molecular identification of the ccs2-1 to ccs2-5 mutations. (A) Schematic representation of the CCS2 immature transcript. Introns are shown in black and exons in gray. The position of the molecular lesions in the ccs2 alleles is indicated. (B) Identified mutations in the ccs2 alleles. The top line corresponds to the nucleotide sequence of the CCS2 ORF and numbers displayed to either side refer to position of the nucleotides within the ORF. The lower-case letters indicate the nucleotide sequence due to change(s) induced by the UV mutagenesis. The lower line shows the corresponding amino acid sequence and the change within the protein sequence resulting from the molecular lesion(s). The asterisk indicates a stop in the protein sequence. The ccs2-1 allele is a CC to TA mutation at position 1036-1037 and the ccs2-3 allele is a G to A change at position 1514. In the ccs2-4, deletion of a single C occurred at position 269. The ccs2-5 mutation combines a G to T mutation at position 826 and deletion of a single G at position 829. The GenBank accession number for the CCS2 nucleotide sequence is KC292647.

or frameshift (ccs2-4 and ccs2-5) mutations, which result in truncations within the first 500 amino acids of the protein except for the ccs2-2 allele (**Figure 3B**). In the ccs2-2 mutant, a single guanine has been deleted from a stretch of 11 guanines in intron 4, 10 bp downstream of the 3<sup>0</sup> end of exon 4. It is possible that this change impairs the splicing of intron 4 in the CCS2 transcript. The identification of the molecular lesions confirms that the complementing sequence we isolated corresponds to the wild-type CCS2 gene rather than an extragenic suppressor of the ccs2 mutation.

The CCS2 gene encodes a 1,719 amino acid protein with a predicted molecular weight of 171,753 Da. The most striking feature of the CCS2 protein is the presence of several 38–40 amino acid repeats occurring between residues 720 and 1610 (**Figure 4AB**). Such motifs, named OPR for **o**ctatrico**p**eptide **r**epeats were first defined in TBC2 (translation factor for chloroplast psbC mRNA), a nuclear encoded protein required for the translation of psbC RNA in the chloroplast (Auchincloss et al., 2002). OPRs are also recognized in other factors controlling translation (TDA1, TAB1), maturation (RAT2, RAA1, RAA8, RAP), or stability (MCG1,MBI1) of chloroplast transcripts (Balczun et al., 2005; Merendino et al., 2006; Eberhard et al., 2011; Rahire et al., 2012; Kleinknecht et al., 2014; Marx et al., 2015; Wang et al., 2015).

The OPR family is further characterized by the presence of low complexity regions, which are regions containing little diversity in their amino acid composition (Balczun et al., 2005; Merendino et al., 2006; Eberhard et al., 2011; Rahire et al., 2012; Marx et al., 2015; Wang et al., 2015). Indeed, we noted the occurrence of several stretches of three or more alanine, glycine, serine, or glutamine repeats in CCS2 (**Figure 4A**). CCS2 has a high content of alanine (24.2%), glycine (10.7%), proline (8.7%), and leucine (8.7%), a feature shared by other OPR proteins (Auchincloss et al., 2002; Balczun et al., 2005; Merendino et al., 2006; Eberhard et al., 2011; Rahire et al., 2012; Marx et al., 2015; Wang et al., 2015). OPR motifs found in CCS2 are highlighted in **Figure 4B** and a consensus OPR motif from all OPR containing proteins that have been functionally identified to date, with the exception of NCC1 and NCC2 (Boulouis et al., 2015), can be seen in **Figure 4C**. The relative locations of the motifs in the proteins used to generate the consensus motif in **Figure 4C** can be seen in **Figure 4B**.

FIGURE 4 | CCS2 is an OPR protein. (A) Amino acid sequence of the deduced CCS2 protein. Light gray text indicates the protein sequence that was fused to GFP for targeting experiments (Figure 6), light gray highlighting indicates OPRs recognized with p-values of at least 1e-10, as shown in (B). The underlined portion indicates the location of the previously recognized "PPPEW" motif. The red arrows indicate the positions of internal HA tags that were introduced to test neutrality of the tag with respect to CCS2 function (see Materials and Methods). (B) MEME-generated distribution of OPR motifs within the Chlamydomonas OPR proteins that have been functionally characterized. Heights of blocks indicate relative proportion of the p-value 1e-10. Red boxes represent motifs found by the MEME program while faded boxes represent motifs found by other programs within the MEME Suite. Sequences of Chlamydomonas OPR proteins CCS2 (KC292647), ASA2 (EDP00850.1), TBC2 (CAD20887.1), TDA1 (CCA62914.20), RAA1 (CAE53330.1), TAB1 (ADY68544.1), RAT2 (EDP02536.1), TAA1 (Cre06.g262650), RAA8 (Cre10.g440000), MCG1 (Cre10.g429400), MBI1 (Cre06.g272450), and Arabidopsis RAP (OAP08625.1) were used (see Supplementary Figure S1). (C) WebLogo consensus sequence of OPRs identified by the MEME program depicted in red in Figure 4B (Bailey and Elkan, 1994; Crooks et al., 2004). Letter height indicates the probability of a particular amino acid (y-axis) at a given position within the 38 amino acid repeat (x-axis). OPR sequences used to calculate this consensus sequence are found in Supplementary Figure S2.

promoter (CCS2-HA↑), respectively. One representative transformant, selected on minimal medium, was used for the analysis. (A) Ten-fold dilution series of algal cells on acetate containing (0.6 µE/m<sup>2</sup> /s) or minimal media (50 µE/m<sup>2</sup> /s), incubated for 1 week at 25◦C. (B) Anti-HA immunodetection using whole cell extracts. The HA-tagged MCA1 strain is used as a control (Raynaud et al., 2007). MCA1 is a plastid-localized PPR protein (PPR for pentatricopeptide repeat). Black arrows indicate the two CCS2 species, gray arrows indicate the two MCA1 species. The lower bands correspond to the expected molecular mass for CCS2 (∼176 kDa) and MCA1 (∼105 kDa). The asterisk indicates the CCS2 species of high molecular weight that is detected in the ccs2 strain expressing CCS2-HA under its native promoter. Ponceau staining serves as a loading control.

### Immunodetection of the CCS2 Protein

In order to detect CCS2 and assess its subcellular localization, a series of constructs expressing HA tagged proteins were engineered. The tagged proteins were either expressed under the control of the native CCS2 promoter or the PSAD promoter, which allows increased expression of Chlamydomonas genes (Fischer and Rochaix, 2001). We saw no difference in the ability of the two HA-tagged CCS2 constructs to complement the ccs2-1 mutation as compared to the WT gene (**Figure 5A**).

Immunoblotting analysis against whole cell extracts from transformants over-expressing CCS2-HA revealed that the protein occurs as two species, one of which appears to migrate at the expected size ('176 kDa) (**Figure 5B**). A similar pattern was also seen in cells expressing HA-tagged MCA1, a PPR (PPR for **p**entatrico**p**eptide **r**epeat) protein with plastid localization (Raynaud et al., 2007; Boulouis et al., 2011). This suggests that the immunoreactive species with higher mobility may be a result of the extraction conditions and/or the denaturation step needed to immunodetect CCS2 (see Materials and Methods). Detection of CCS2-HA was only possible from freshly grown cultures when a cocktail of protease inhibitors was used immediately followed by denaturing at 70◦C (instead of 100◦C), an indication that the protein is likely highly sensitive to proteolysis. Interestingly, Rahire et al. (2012) noted that the OPR protein TDA1 was also very susceptible to proteolysis.

## The CCS2 Protein Localizes to the Plastid

A chloroplast targeting sequence is assigned by both ChloroP (Emanuelsson et al., 1999) and WoLF PSORT (Horton et al., 2007) at the N-terminus of CCS2, an indication that CCS2 might reside in the plastid. However, the extraction methods necessary for detection of CCS2-HA precluded the use of subcellular fractionation to determine protein localization. Hence, we tested the ability of the CCS2 N-terminus to direct GFP to the plastid in a heterologous system such as N. benthamiana. This is justified, as targeting sequences from Chlamydomonas nuclear-encoded proteins retain their function as transit peptides for import into chloroplasts in several species of land plants including Nicotiana (Nakazato et al., 2003; Falciatore et al., 2005; Levitan et al., 2005; Li et al., 2016; Yamaoka et al., 2016). To this end, we constructed CCS2-GFP, which encoded a protein consisting of the first 100 amino acids of CCS2, including the putative targeting sequence, translationally fused to the amino-terminus of GFP. This CCS2-GFP expressing construct was introduced into N. benthamiana leaves. Fluorescence microscopy shows clear overlay of chlorophyll auto-fluorescence, which is a feature of the chloroplast, and GFP fluorescence (**Figure 6**). These results indicate that the first 100 amino acids of CCS2 are sufficient to target GFP to the chloroplast of N. benthamiana and hence the CCS2 protein is likely localized to the plastid of Chlamydomonas.

### DISCUSSION

### Identification of CCS2 as an OPR Protein Controlling Cytochrome c Assembly

Here, we have identified CCS2, a chloroplast localized cytochrome c assembly factor that, based on primary sequence similarity, appears to be unique to specific branches of the chlorophycean algae. This protein has been recognized as a member of the recently designated OPR family, which contains 43 members in Chlamydomonas (Eberhard et al., 2011). OPRs are defined by loosely conserved repeats of 38–40 amino acids

(Auchincloss et al., 2002; Eberhard et al., 2011; Rahire et al., 2012).

Via bioinformatics analysis, OPR proteins have been assigned to the 'α-solenoid' superfamily, which contains proteins defined by similar tandem repeats, such as TPRs (tetratricopeptide repeat, 34 amino acids) and PPRs (pentatricopeptide repeat, 35 amino acids) (Eberhard et al., 2011; Rahire et al., 2012). Three dimensional structure analyses of TPRs (Das et al., 1998; D'Andrea and Regan, 2003; Zeytuni and Zarivach, 2012) and PPRs (Ringel et al., 2011; Nakamura et al., 2012; Shen et al., 2016) indicate that these motifs result in a series of anti-parallel α-helices. Protein predictions using I-TASSER (Zhang, 2008; Roy et al., 2010) suggest that the OPR proteins are organized in a similar manner (Auchincloss et al., 2002; Eberhard et al., 2011; Rahire et al., 2012). I-TASSER also predicts regions of anti-parallel α-helices in CCS2 (not shown). TPR proteins act as scaffolds to mediate protein–protein interactions and control a wide range of cellular functions such as the cell cycle, transcription in the nucleus, or protein import into mitochondria and peroxisomes (D'Andrea and Regan, 2003). On the other hand, all the PPR proteins described so far were shown to be control maturation, stability, or translation of organellar RNAs, presumably via direct interaction with their target transcript(s) (Barkan and Small, 2014; Manna, 2015).

The PPR family has expanded greatly in the plant lineage, with 450 representatives in Arabidopsis alone, while the typical non-plant, eukaryotic genome encodes for fewer than 40 representatives (Barkan and Small, 2014; Manna, 2015). Similar to the PPR protein family expansion seen in land plants, chlorophycean algae have an expansion in the number of OPR containing proteins (Eberhard et al., 2011; Rahire et al., 2012). The majority of functionally characterized OPR proteins are involved in translation, maturation, or stability of transcripts in the chloroplast of Chlamydomonas. For instance, TBC2, TDA1, and TAB1 are factors involved in the translation of the psbC, atpA, or psaB transcripts, respectively (Auchincloss et al., 2002; Eberhard et al., 2011; Rahire et al., 2012), while RAA1/RAA8 and RAT2 are involved in the maturation of psaA and tscA RNAs (Balczun et al., 2005; Merendino et al., 2006; Marx et al., 2015). In Arabidopsis, RAP is the sole OPR containing protein and loss of RAP function yields a defect in the maturation of the chloroplast 16S ribosomal RNA (Kleinknecht et al., 2014). Recent studies have uncovered the role of MCG1 and MBI1 in stabilizing the petG mRNA, encoding a small subunit of the cytochrome b6f complex and the psbI mRNA, coding for a small subunit of photosystem II, respectively (Wang et al., 2015).

If the role of most OPR containing proteins as transcriptinteracting factors was inferred from the phenotypic analysis of loss-of-function mutations, the demonstration that OPR motifs interact directly with their relevant target RNAs was only provided for TAB1 (Rahire et al., 2012). A recent study revealed that gain of function mutations in the OPR motifs of NCC1 and NCC2 confer an ability to recognize chloroplast transcripts as novel targets of action (Boulouis et al., 2015). This suggests that specific amino acids within these OPR motifs must govern nucleotide recognition, but the basis of this specificity still remains to be deciphered.

The molecular mass range of the functionally identified OPR proteins extends from 45 to 269 kDa with an average length of 1428 amino acids and 5–17 OPR motifs. These repeats are found primarily in the central regions of proteins while the carboxyland amino-termini are typically characterized by stretches of single amino acid repeats (Auchincloss et al., 2002; Merendino et al., 2006; Rahire et al., 2012). The OPR motifs themselves can be highly divergent, and so the repeat count per protein fluctuates depending on how stringently one defines a single OPR motif

(**Figure 4B**). This divergence is reflected in the fact that OPR repeats in CCS2 are characterized by a LWALAR consensus motif (**Figure 4A**) while repeats in TBC2, TDA1, and TAB1 contain a PPPEW sequence (Auchincloss et al., 2002; Eberhard et al., 2011; Rahire et al., 2012). While in some instances OPR motifs are serially repeated, the motifs can also be separated by gaps (**Figure 4B**), often by stretches of amino acids such as alanine. We noted the frequent occurrence of a poly-glutamine stretch upstream of the OPR motif in CCS2. Interestingly, glutamine-rich regions are known to form polar zippers involved in protein interaction but they have also been recognized as motifs for transcriptional activation (Courey and Tjian, 1988; Stott et al., 1995).

That CCS2 functions directly in the heme attachment reaction in the thylakoid lumen is very unlikely, considering its large size and the absence of a typical bipartite targeting sequence for luminal localization. Considering its inclusion in the OPR family, one possibility is that CCS2 controls the maturation, stability, or translation of a chloroplast transcript in the stroma involved in cytochrome c assembly. In the chloroplast genome, the ccsA gene encodes a heme delivery factor required for cytochrome c maturation (Xie and Merchant, 1996). While the level of ccsA transcript in the ccs2 mutants was previously shown to be unaltered, indicating that CCS2 is not required to stabilize the ccsA mRNA (Xie et al., 1998), the hypothesis that CCS2 is involved in the translation of ccsA remains plausible. Unfortunately, this was not tested due to the lack of an anti-CcsA antibody. Another likely scenario is that CCS2 functions like a TPR containing protein and mediates the assembly of protein complexes containing other CCS factors. The OPR family also includes ASA2 (ATP Synthase Associated protein), a subunit of the unusual mitochondrial ATP synthase, found only in chlorophycean algae (van Lis et al., 2007; Cano-Estrada et al., 2010) (**Figure 4B**). While a 200 kDa CCS complex containing ccsA (40 kDa) and Ccs1 (60 kDa) has been identified at the thylakoid membrane (Hamel et al., 2003), it is unlikely that the

### REFERENCES


170 kDa CCS2 is a component of this complex because of its large size. However, it is conceivable that CCS2 could stabilize the CCS complex via protein-protein interactions or facilitate the recruitment of the components of this complex on the stromal side of the thylakoid membrane.

### AUTHOR CONTRIBUTIONS

SC executed the experiments, interpreted the data and wrote the manuscript; PH designed the experiments, interpreted the data and wrote the manuscript; IL executed experiments.

### FUNDING

This work was primarily supported by a National Science Foundation grant (MCB-0920062) and a U.S. Department of Energy (DOE), Office of Science, Basic Energy Sciences (BES) grant (DE-SC0014562) to PH.

### ACKNOWLEDGMENTS

We thank Dr. Meier for the gift of the GFP-NLS-GFP-NES expressing construct and Dr. Zhou for technical help with the microscopy experiments. We also wish to thank Dr. O. Vallon for scientific discussion on the OPR proteins and Dr. R. Lamb for her comments on the manuscript. PH thanks Sabeeha Merchant in whose laboratory the project was initiated.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpls.2017.01306/ full#supplementary-material


Polytomella sp. Biochim. Biophys. Acta 1797, 1439–1448. doi: 10.1016/j.bbabio. 2010.02.024



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Cline, Laughbaum and Hamel. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# GUN1, a Jack-Of-All-Trades in Chloroplast Protein Homeostasis and Signaling

Monica Colombo<sup>1</sup> , Luca Tadini <sup>2</sup> , Carlotta Peracchio<sup>2</sup> , Roberto Ferrari <sup>2</sup> and Paolo Pesaresi <sup>2</sup> \*

<sup>1</sup> Centro Ricerca e Innovazione, Fondazione Edmund Mach, San Michele all'Adige, Italy, <sup>2</sup> Dipartimento di Bioscienze, Università degli Studi di Milano, Milan, Italy

The GENOMES UNCOUPLED 1 (GUN1) gene has been reported to encode a chloroplast-localized pentatricopeptide-repeat protein, which acts to integrate multiple indicators of plastid developmental stage and altered plastid function, as part of chloroplast-to-nucleus retrograde communication. However, the molecular mechanisms underlying signal integration by GUN1 have remained elusive, up until the recent identification of a set of GUN1-interacting proteins, by co-immunoprecipitation and mass-spectrometric analyses, as well as protein–protein interaction assays. Here, we review the molecular functions of the different GUN1 partners and propose a major role for GUN1 as coordinator of chloroplast translation, protein import, and protein degradation. This regulatory role is implemented through proteins that, in most cases, are part of multimeric protein complexes and whose precise functions vary depending on their association states. Within this framework, GUN1 may act as a platform to promote specific functions by bringing the interacting enzymes into close proximity with their substrates, or may inhibit processes by sequestering particular pools of specific interactors. Furthermore, the interactions of GUN1 with enzymes of the tetrapyrrole biosynthesis (TPB) pathway support the involvement of tetrapyrroles as signaling molecules in retrograde communication.

Keywords: nucleoid, GUN1, protein homeostasis, retrograde signaling, biogenic control

### INTRODUCTION

Upon illumination, proplastids differentiate into functional chloroplasts in developing photosynthetic tissues of cotyledons, leaves, and stems (Jarvis and López-Juez, 2013). Chloroplast biogenesis also occurs during the growth of young green tissues, as cells expand and mature chloroplasts undergo division by binary fission (Okazaki et al., 2010). This process is characterized macroscopically by rapid greening of the young chloroplast and microscopically by the concomitant formation of thylakoid membranes and the reorganization of nucleoids, i.e., DNAcontaining structures without defined boundaries, which differ in number, size, and distribution within plastids at different developmental stages, and harbor the plastid gene expression (PGE) machinery (Pfalz and Pfannschmidt, 2013; Melonek et al., 2016).

At the molecular level, this rather complex biogenic transition is achieved by cytosolic synthesis of chloroplast-targeted proteins, followed by import, assembly, folding, and degradation of unfolded/misfolded proteins (Jarvis and López-Juez, 2013). Indeed, the plastid genome itself (the

#### Edited by:

Fiammetta Alagna, Research Council for Agriculture and Agricultural Economy, Italy

#### Reviewed by:

Lynn Richardson, Michigan State University, USA Silvia Ramundo, University of California, San Francisco & Carnegie Institution for Science & Howard Hughes Medical Institute, USA

\*Correspondence:

Paolo Pesaresi paolo.pesaresi@unimi.it

#### Specialty section:

This article was submitted to Plant Cell Biology, a section of the journal Frontiers in Plant Science

Received: 21 July 2016 Accepted: 07 September 2016 Published: 22 September 2016

#### Citation:

Colombo M, Tadini L, Peracchio C, Ferrari R and Pesaresi P (2016) GUN1, a Jack-Of-All-Trades in Chloroplast Protein Homeostasis and Signaling. Front. Plant Sci. 7:1427. doi: 10.3389/fpls.2016.01427 plastome) comprises fewer than 100 protein-coding genes, and the vast majority of the 2000–3000 proteins that make up the chloroplast proteome are encoded in the nucleus (Richly and Leister, 2004). In particular, precursor proteins carrying N-terminal transit peptides initially interact with two multiprotein complexes termed Translocon at the outer envelope membrane of chloroplasts (TOC) and Translocon at the inner envelope membrane of chloroplasts (TIC), which facilitate their active transport through the chloroplast envelope, powered by an ATP import motor, consisting of the stromal heatshock protein 93 (Hsp93), heat-shock protein 70 (Hsp70), and heat-shock protein 90 (Hsp90; Flores-Perez and Jarvis, 2013; Inoue et al., 2013; Shi and Theg, 2013a,b). Upon translocation, proteins are exposed to different proteolytic systems of prokaryotic origin, which are responsible for protein maturation, control of protein abundance, and removal of either misfolded or damaged components. Among these, the stromal protease Clp is a multimeric complex made of chaperones and serine protease subunits, which serve general housekeeping functions. In contrast, the thylakoid-associated FtsH (Filamentous temperature sensitive H) proteases are zinccontaining metalloendopeptidases that have both chaperone and proteolytic functions, and participate in the Photosystem II repair cycle, together with the DEG serine proteases (Kato and Sakamoto, 2010; Van Wijk, 2015).

Besides translation and post-translational processes, chloroplast biogenesis also requires transcriptional coordination of thousands of nuclear genes with the expression of the comparatively few plastid genes in order to meet the needs of the developing chloroplast (Chan et al., 2016; Kleine and Leister, 2016). This is achieved through extensive exchange of information between plastids and the nucleus, for instance, via biogenic retrograde signaling—a system in which developmentally relevant stimuli in plastids induce the accumulation of specific signaling molecules that relay information to the nucleus, and in turn adjust the expression of nuclear genes to the needs of the plastids (Pogson et al., 2008; Woodson and Chory, 2008; Chan et al., 2016).

During the last 30 years, experiments with the carotenoid biosynthesis inhibitor norfluorazon (NF) and the inhibitor of plastid translation lincomycin (LIN), each of which arrests chloroplast development at the proplastid stage and represses the expression of photosynthesis-associated nuclear genes (PhANGs), have provided insights into the plastid's biogenic retrograde pathways (Oelmüller and Mohr, 1986; Oelmüller et al., 1986).

Six genome uncoupled (gun) mutants have been characterized in Arabidopsis thaliana that fail to repress transcription of the nuclear gene Lhcb1.2 after NF treatment, and are thus impaired in retrograde signaling (Susek et al., 1993; Mochizuki et al., 2001; Larkin et al., 2003; Koussevitzky et al., 2007; Adhikari et al., 2011; Woodson et al., 2011). Five of these genes, GUN2-6, were found to be involved in tetrapyrrole biosynthesis (TPB), whereas GUN1, which encodes a nucleoid-localized pentatricopeptide repeat protein (PPR), has been shown to have a role in PGE, and to act as an integrator of multiple retrograde signals, since gun1 mutants are unique in exhibiting a gun phenotype in response to both norfluorazon and lincomycin (Gray et al., 2003; Koussevitzky et al., 2007). However, the exact molecular role of GUN1 remained enigmatic until the new insights provided by the recent identification of a set of GUN1-interacting proteins (Tadini et al., 2016; **Table 1**).

Based on the functions of these partners, GUN1 appears to take part in multiple processes essential for chloroplast biogenesis and maintenance of the chloroplast proteome. GUN1 mediated control of plastid ribosomal protein S1 (PRPS1) accumulation, together with co-immunoprecipitation (CoIP) of proteins involved in different steps of plastid translation, support the involvement of GUN1 in the regulation of plastid protein synthesis. Furthermore, the presence of several chaperones in the CoIP mixture suggests a role for GUN1 in the coordination of chloroplast protein import and protein degradation.

Intriguingly, several GUN1 interactors appear to accumulate to higher levels upon induction of the unfolded protein response (UPR) in Chlamydomonas reinhardtii chloroplasts, which is triggered upon conditional repression of the catalytic subunit of Clp protease (ClpP1; Ramundo et al., 2013; Ramundo and Rochaix, 2014; Rochaix and Ramundo, 2015). This finding suggests the possible involvement of GUN1 in the UPR signaling pathway.

In this review, we describe the functional roles of the different GUN1 protein partners and propose some testable hypotheses that should clarify the molecular role of GUN1 in chloroplast biogenesis and chloroplast protein homeostasis.

### GUN1 IS FOUND IN PLASTID NUCLEOIDS AND INTERACTS WITH THE TRANSCRIPTIONAL MACHINERY

GUN1 encodes a member of PPR-containing protein family, which has a small MutS-related (SMR) domain at the C-terminal end and a plastid targeting signal sequence at its N terminus. PPR motifs have been shown to mediate interactions with nucleic acids, and the SMR domain is found in proteins that act in DNA repair and recombination. However, in vivo RNA and DNA immunoprecipitation on chip (NIP-chip), as well as one-hybrid assays, have failed to detect any stable interaction of GUN1 with nucleic acids (Tadini et al., 2016), in contrast to a previous report, in which a GUN1 fragment containing both the PPR and SMR domains was shown to bind DNA in vitro (Koussevitzky et al., 2007). Nevertheless, GUN1 appears to be associated with nucleoids in the chloroplast, and more specifically with the domain of active plastid transcription, as shown by the relatively large and distinct foci of a fluorescent GUN1-YFP (Yellow Fluorescence Protein) chimera that co-localize with a Plastid Transcriptionally Active Chromosome 2-Cyan Fluorescence Protein (pTAC2-CFP) fusion in chloroplasts of mesophyll cells (Koussevitzky et al., 2007). However, although the repertoire of nucleoid-associated proteins so far identified is quite extensive, the GUN1 protein is not listed in any of the chloroplast or nucleoid/pTAC proteomes published to date (for a review see Melonek et al., 2016), most probably because it accumulates in very small amounts at specific developmental stages or under

#### TABLE 1 | GUN1 interactors together with their functions and impacts on plant development.


(Continued)

#### TABLE 1 | Continued


Note that proteins Q9SIP7 (AT2G31610) and Q42112 (AT3G09200) reported to be identified in coimmunoprecipitates of GUN1-GFP (Tadini et al., 2016) are not listed in this Table, since they have been described as subunits of cytosolic ribosomes. Furthermore, the protein Q9C5C2 (AT5G25980) has not been included, since it localizes to the tonoplast (Agee et al., 2010).

n.d., not determined.

<sup>a</sup>Phenotype of knock-out mutants is described.

<sup>b</sup>Protein already identified as part of chloroplast nucleoid by proteomic approaches.

<sup>c</sup>Assays used to identify the corresponding protein as a GUN1 interactor: coimmunoprecipitation followed by mass spectrometry (CoIP-MS), yeast two-hybrid (Y2H) analysis, and Bimolecular Fluorescence Complementation (BiFC).

particular physiological conditions. This inference is supported by CoIP experiments with a Green Fluorescence Protein (GUN1- GFP) fusion and subsequent mass spectrometry (MS), which identified several nucleoid subunits as interactors with GUN1 (Tadini et al., 2016; **Table 1**).

pTAC6 is among the GUN1 interactors, and it has been reported to interact directly with the plastid-encoded RNA polymerase (PEP), building together with pTAC2 and other polymerase-associated proteins (PAPs) the soluble RNA polymerase (sRNPase) complex (Pfalz et al., 2006), a central component of nucleoids (Steiner et al., 2011; **Figure 1**). Intriguingly, pTAC6 (also known as PAP8) contains no known domain and exhibits no homologies that could provide hints as to its function in PGE (Steiner et al., 2011). However, functional genomics analyses have indicated that homozygous pap knockout lines develop white cotyledons, fail to accumulate chlorophyll even under low light intensities, and do not produce primary leaves unless they are cultivated on MS medium supplemented with sucrose (for a review, see Pfalz and Pfannschmidt, 2013). Furthermore, analyses of PGE in pap mutants revealed strong repression of the accumulation of PEP-dependent transcripts, whereas levels of nucleus-encoded RNA polymerase (NEP) dependent transcripts were not depleted, while some were enhanced, indicating that pTAC6/PAP8 and the other PAP proteins are essential for the activity of PEP (see **Table 1**).

### GUN1 CONTROLS PLASTID TRANSLATION AND RIBOSOME BIOGENESIS

GUN1 also interacts with several ribosomal subunits, such as the plastid-encoded ribosomal proteins L2, S3, and S4 (rpl2, rps3, and rps4) and the nucleus-encoded plastid ribosomal protein L10 (PRPL10; **Figure 1**). Furthermore, yeast two-hybrid and Bimolecular Fluorescence Complementation (BiFC) assays revealed a physical interaction between GUN1 and PRPS1 (Tadini et al., 2016). Ribosomal proteins have been reproducibly detected in nucleoid and pTAC proteomes (Melonek et al., 2016), further supporting the existence of a translational subdomain within the nucleoids, as proposed by Pfalz and Pfannschmidt (2013). The homologs of PRPL10, rpl2, PRPS1, rps3, and rps4 are essential components of the protein biosynthetic machinery in Escherichia coli (Baba et al., 2006; Shoji et al., 2011) and the indispensability of rps3 and rps4 has been also proven in tobacco plastids (Rogalski et al., 2008; Fleischmann et al., 2011). Furthermore, PRPL10 is annotated as EMBryo defective 3136 (EMB3136) in the SeedGenes Project database (http://www.seedgenes.org/), and in its absence Arabidopsis embryo development arrests at the globular stage (Bryant et al., 2011). Mutants devoid of PRPS1 have not been described. However, given the conservation of PRPS1 function in prokaryotes and chloroplasts, it can be confidently assumed that complete lack of PRPS1 is lethal in Arabidopsis.

Taking into consideration the function of these ribosomal proteins, it can be argued that their interaction with GUN1 has a dual purpose. On the one hand, GUN1 modulates protein synthesis by controlling the abundance of PRPS1, which, together with rps3 and rps2, has been reported to form the domain responsible for the interaction of the 30S ribosomal subunit with mRNA, promoting translation initiation (Manuell et al., 2007; Tadini et al., 2016). This role is supported further by the stable interaction of GUN1 with the chloroplast translation initiation

factor 2 (cpIF2; Tadini et al., 2016), also known as FUG1, and reported to be essential for chloroplast biogenesis (Miura et al., 2007).

On the other hand, GUN1 seems to be involved in the process of ribosome biogenesis too, since nucleoid-associated ribosomes are thought to be in various stages of assembly, with several rRNA maturation steps occurring in a co-transcriptional and assembly-assisted manner, as in prokaryotic systems (Bohne, 2014). For instance, the DEAD-box-containing, ATP-dependent RNA helicase 3 (RH3), which has been functionally linked to the chloroplast nucleoid (Majeran et al., 2012), is among the proteins that interact with GUN1 (Tadini et al., 2016; see also **Figure 1** and **Table 1**). RH3 is directly involved in the splicing of group II introns in the rpl2, trnA, trnI, and rps12 transcripts and could be coimmunoprecipitated with immature and mature 23S rRNA (Asakura et al., 2012). Furthermore, the PPR protein At3g49240 also known as AtPPR\_3g49240, according to the PPR protein database (http://www.plantenergy.uwa.edu.au/applications/ppr/ppr.php), is also part of GUN1's interactors, and its maize ortholog, GRMZM2G074599\_P01, has been identified in the chloroplast nucleoid (Majeran et al., 2012). The gene is annotated as embryo defective 1796 (EMB1796) in the SeedGenes database, since the complete lack of AtPPR\_3g49240 leads to the arrest of embryonic development at the globular stage (Cushing et al., 2005), further supporting the essential role of GUN1 interactors in chloroplast biogenesis.

### GUN1 AND THE IMPORT OF CHLOROPLAST PROTEINS

Almost a quarter of the GUN1 interactors identified by CoIP-MS are chaperones (see **Table 1**), a relatively high proportion when compared with the extensive repertoire of protein functions found within the nucleoid (Melonek et al., 2016). The stromal Hsp93 and Hsp70 chaperones mediate different steps in protein import into the chloroplast stroma, whereas the 60 KD chaperonin Cpn60 is thought to be involved in the folding of newly imported mature proteins and to function downstream of Hsp93 and Hsp70 (Kessler and Blobel, 1996; Jackson-Constan et al., 2001; Flores-Perez and Jarvis, 2013). Furthermore, the two genes most highly co-regulated with GUN1 encode the proteins TIC110 and TOC159 (Tadini et al., 2016), which are part of the outer and inner chloroplast translocons, respectively, suggesting a role of GUN1 in chloroplast protein import (**Figure 2**).

### The Hsp93 Chaperones

In Arabidopsis, there are two nearly identical isoforms of Hsp93, termed Hsp93-V and Hsp93-III (or ClpC1 and ClpC2, respectively) and both interact with GUN1. The two proteins are highly homologous, but Hsp93-V is expressed at much higher levels than Hsp93-III (Kovacheva et al., 2005, 2007), and only Hsp93-V has been reported as a component of the nucleoid proteome (Phinney and Thelen, 2005; Majeran et al., 2012; Melonek et al., 2012; Huang et al., 2013). Furthermore, both hsp93 single mutants are viable whereas hsp93-III hsp93-V double mutant is embryo-lethal, indicating that the two proteins have redundant functions in Arabidopsis chloroplasts (Constan et al., 2004; Sjögren et al., 2004; Kovacheva et al., 2007).

The current model for chloroplast protein import assumes that the preprotein transit peptide interacts with the TOC, and is subsequently transported through the TIC in an energydependent process (Shi and Theg, 2013b). In particular, the Tic110–Tic40 interaction is proposed to trigger the release of the transit peptide from Tic110 and enable the association of the preprotein with Hsp93 (Inaba et al., 2003). Tic40 then stimulates ATP hydrolysis by Hsp93, which harnesses the energy released to draw the preprotein into the stroma (Chou et al., 2006).

### The Hsp70 Chaperones

Recent work has also demonstrated the involvement of Hsp70 in protein translocation into chloroplasts, as part of the translocon energy-dependent engine together with Hsp93 and Hsp90 (Inoue et al., 2013; Liu et al., 2014). Like Hsp93, Hsp70 proteins occur in two isoforms, Hsp70-1 and Hsp70-2, in the chloroplasts of Arabidopsis (Su and Li, 2008) and only Hsp70-1 was found in the proteomes of pTAC and crude nucleoids (for a review see Melonek et al., 2016). However, both Hsp70 proteins have been identified as GUN1 interactors (Tadini et al., 2016). Protein import assays using chloroplasts isolated from the Arabidopsis Hsp70 knockout mutants hsp70-1 and hsp70- 2 showed that stromal Hsp70s are important for the import of both photosynthetic and non-photosynthetic precursor proteins, especially in early developmental stages (Su and Li, 2010). Furthermore, no hsp70-1 hsp70-2 double mutant has ever been isolated. Thus, the two Hsp70s are likely to have redundant functions that are essential for plant development and chloroplast biogenesis.

### The Cpn60 Chaperonins

After preproteins delivered to the stroma have been processed, they may require accessory factors to enable them to fold into their functional conformation, or to reach their final intra-organellar destination. The stromal molecular chaperones Hsp70, Cpn60, and Cpn10 are all believed to mediate the folding or onward guidance of newly imported polypeptide chains (Boston et al., 1996; Jackson-Constan et al., 2001). In particular, immunoprecipitation experiments have revealed that Cpn60 operates in close proximity with Tic110 (Kessler and Blobel, 1996), while import experiments have shown a transient association of mature, newly imported proteins with the Cpn60- Tic110 complex, suggesting that Tic110 can recruit Cpn60 in an ATP-dependent manner for the folding of proteins upon their arrival in the stroma. It has also been suggested that stromal Hsp70 and Cpn60 act sequentially to facilitate the maturation of imported proteins, particularly those destined for the thylakoid membranes (Madueno et al., 1993; Tsugeki and Nishimura, 1993; Peng et al., 2011). The Arabidopsis genome encodes two members of the Cpn60α family, denoted ptCpn60α1 and ptCpn60α2, and four members of Cpn60β, known as ptCpn60β1–β4 (Suzuki et al., 2009). Two of them, ptCpn60α1 and ptCpn60β2, have been linked to the nucleoid proteome (Melonek et al., 2016), and ptCpn60α1 and ptCpn60β1 are among the GUN1 interactors identified via the CoIP-MS strategy (see **Table 1**). The complete loss of ptCpn60α1, in the

mutant termed schlepperless (slp), causes retardation of embryo development before the heart stage and an albino seedling phenotype, indicating that ptCpn60α1 is essential for chloroplast biogenesis (Apuya et al., 2001). Conversely, plants devoid of ptCpn60β1, also known as lesion initiation 1 (len1), have leaves with wrinkled and irregular surfaces and undergo localized, spontaneous cell death in the absence of pathogen attack, i.e., lesion formation, under short-day conditions (Ishikawa et al., 2003).

### OTHER FUNCTIONS OF PLASTID CHAPERONES

Besides their roles in plastid protein import, all GUN1 interacting chaperones are present in the stroma at significant amounts relative to their association with the chloroplast import apparatus and perform various other functions together with different protein complexes (**Figure 2**). For instance Hsp93, also termed ClpC, acts as a regulatory chaperone in the Clp protease complex, the most abundant stromal protease with general household functions (Sakamoto, 2006; Van Wijk, 2015). Clp substrates are selected through various signals intrinsic to amino acid sequences and the ATP-dependent ClpC chaperone activity helps to progressively unfold selected substrates that are delivered to the ClpPR core for degradation into small peptides (∼8–10 amino acids long; Olinares et al., 2011).

Similarly, Cpn60 forms a large oligomeric protein complex (>600 KDa) that promotes the assembly of Rubisco (Gutteridge and Gatenby, 1995). In particular, it has been observed that the large subunit of Rubisco (RbcL) is specifically associated with Cpn60 before assembly into the holoenzyme and that the Cpn60- RbcL complex is an obligatory intermediate. Furthermore, Cpn60 proteins have been shown to be essential for plastid division in A. thaliana (Suzuki et al., 2009). Thus, mesophyll cells in ptcpn60α1-1 (a missense mutant) and ptcpn60β1-1 (a protein null) plants, contain fewer and larger chloroplasts, indicating that normal levels of plastid Cpn60 are required for the correct folding of the stromal plastid division proteins and/or regulation of FtsZ (Filamentous temperature-sensitive Z) polymer dynamics (Suzuki et al., 2009).

The same holds true for the Hsp70 proteins, which are also involved in modulation of protein activity, regulation of protein degradation and prevention of irreversible protein aggregation when they are free in the stroma (Su and Li, 2008). Potentially GUN1 can be involved in a multitude of activities, besides plastid protein import, thus further investigations are needed to clarify the functional significance of GUN1–chaperone interactions.

### GUN1 AND THE CHLOROPLAST UNFOLDED PROTEIN RESPONSE (cpUPR)

Chaperones, together with enzymes that process and degrade proteins, are also necessary to maintain protein folding homeostasis in the various compartments of eukaryotic cells. Distinct signal transduction pathways, known as unfolded protein responses (UPRs), have evolved to couple the unfolded/misfolded protein load to the expression of specific chaperones and enzymes that promote folding and the disposal of misfolded proteins in each compartment.

The unfolded protein response was first discovered in the endoplasmic reticulum (ER) in yeast, where inhibition of protein folding leads to the transcriptional up-regulation of several chaperones (Cox et al., 1993), and subsequently in mitochondria, where accumulation of unfolded proteins in the mitochondrial matrix stimulates the expression of nuclear gene transcripts coding for mitochondrial chaperones (Aldridge et al., 2007; Lin and Haynes, 2016). Compared to yeast and metazoans, studies of plant UPRs are less advanced, and molecular details are known mainly for the ER-dependent UPR, which shows certain similarities with the process in multicellular eukaryotes, as well as plant-specific features (Ruberti et al., 2015). Recently, the possible existence of a chloroplast UPR (cpUPR) has been investigated in the green alga Chlamydomonas reinhardtii. Taking advantage of a repressible chloroplast gene expression system (Rochaix et al., 2014), Ramundo et al. (2014) induced the selective gradual depletion of the essential stromal Clp protease, in order to follow the early and late events caused by the decrease in its abundance. Temporal profiles of gene expression and protein accumulation revealed a marked increase in levels of chaperones, including Hsp70B, upon Clp depletion. Similar data have also been reported for Arabidopsis, where up-regulation of chloroplast chaperones and protein-sorting components occurred upon constitutive repression of Clp (Rudella et al., 2006; Zybailov et al., 2009). In particular, characterization of total leaf proteomes of WT and clpr2-1 highlighted differential expression of 768 proteins. The largest functional category quantified (with 205 proteins) comprised proteins involved in translation, folding and degradation. Strikingly, all the chaperones interacting with GUN1, including Hsp93, Hsp70, Cpn60, as well as the DEAD box RNA helicase RH3, are among those up-regulated (by between 1.6- and 8.5-fold) in clpr2-1 leaves, whereas no significant change in the chloroplast ribosomal protein population was observed (Zybailov et al., 2009).

Taken together, these findings suggest that disruption of protein homeostasis in organelles can be sensed and transduced to the nucleus to induce the expression of a specific set of factors responsible for promoting folding and monitoring protein quality control (Ramundo and Rochaix, 2014; Rochaix and Ramundo, 2015). After entering the higher plant chloroplast, these factors are able to interact with the nucleoid-associated GUN1 protein (**Figure 2**), which might therefore play a role in the cpUPR process.

### GUN1 AND CHLOROPLAST METABOLISM

The large subunit of ribulose bisphosphate carboxylase (RbcL) and the β subunit of the ATP synthase are also among the interactors of GUN1 identified by CoIP-MS analysis (Tadini et al., 2016). Because of their relatively high abundance in the chloroplast proteome, it is tempting to assume that these proteins are simply contaminants. However, RbcL and subunits of the ATP synthase have been repeatedly identified in the pTAC/nucleoid proteomes, even though different procedures were employed for isolation of crude nucleoid fractions and highly purified pTAC complexes (for a review see Melonek et al., 2016), thus suggesting these proteins might have a dual localization to the chloroplast stroma and nucleoids. The nucleoid association of RbcL and ATP synthase, i.e., of proteins that are not directly involved in core nucleoid functions, might also indicate that nucleoids also monitor photosynthesis and energy metabolism and respond appropriately to any perturbations (**Figure 3**).

Unlike RbcL and the ATP synthase β subunit, RETICULATA-RELATED 4 (RER4), an integral component of the chloroplast envelope membranes with three transmembrane α-helices, has never been identified in the pTAC/nucleoid proteome, although it appears to be an interactor of GUN1 (**Table 1**). The mutant rer4-1 exhibits leaf reticulation, having green veins that stand out against paler intervein tissue, with fewer and smaller mesophyll cells than those of the wild type leaves (Perez-Perez et al., 2013). The molecular function of RER4 remains to be established. However, some hints as to its role in the chloroplast can be derived from features of the rer4- 1 mutant phenotype. A possible involvement of RER4 in retrograde signaling is suggested by the altered growth and development of mesophyll cells. Alternatively, the absence of RER4 might deplete the supply of essential metabolites during early stages of leaf development, which could explain the aberrant mesophyll structure. Furthermore, RER4 has been suggested to be involved in the control of reactive oxygen species (ROS), since the reticulated pigmentation of the rer4- 1 mutant grown under long-day conditions can be rescued

by a short-day photoperiod, which markedly dampens ROS accumulation.

The 2-Cys peroxiredoxin A (2-Cys Prx A; see also **Table 1**), another interactor with GUN1, appears also to have a role in ROS scavenging (Rey et al., 2007; Pulido et al., 2010; Dietz, 2016) and, like RER4, it has never been reported to be part of the pTAC/nucleoid proteome (Pfalz et al., 2006; Majeran et al., 2012; Huang et al., 2013). 2-Cys Prx A and the highly homologous 2-Cys Prx B function as redox sensors and chaperones, thanks to the flexibility of their protein structure (König et al., 2013), and they have been shown to control the conversion of Mg-protoporphyrin monomethyl ester into protochlorophyllide (Stenbaek et al., 2008).

The involvement of GUN1 in TPB is further supported by its interaction with four TPB enzymes, namely subunit D of Mg chelatase (CHLD), porphobilinogen deaminase (PBGD), uroporphyrinogen III decarboxylase (UROD2), and ferrochelatase I (FC1), as demonstrated by both yeast two-hybrid and BiFC assays (Tadini et al., 2016; **Figure 3**). Interestingly, mutants defective in three of these GUN1 interactors—CHLD, PBGD, and FC1—have themselves been described as gun mutants (Strand et al., 2003; Huang and Li, 2009; Woodson et al., 2011), but have never been identified in crude nucleoid preparations, unlike subunit I of Mg chelatase (CHLI; Melonek et al., 2012; Huang et al., 2013).

### GUN1 AND PLASTID PROTEIN HOMEOSTASIS: SOME TESTABLE HYPOTHESES

The recent identification of the GUN1 protein's partners in chloroplasts of Arabidopsis by means of CoIP-MS studies as well as in yeast two-hybrid and BiFC assays (Tadini et al., 2016) strongly suggests a major role for GUN1 in plastid protein homeostasis (**Figure 4**). This regulatory role involves proteins that are, in most cases, members of multimeric protein complexes and whose functions are often context-dependent. Furthermore, most GUN1 interactors appear to participate in four major processes:

### Chloroplast Protein Synthesis

A wealth of evidence accumulated during the last two decades supports the primacy of plastid protein synthesis in the control of chloroplast gene expression (Choquet and Wollman, 2002; Manuell et al., 2007; Tiller and Bock, 2014; Sun and Zerges, 2015). In this context, GUN1 has been suggested to regulate translation in plastids by modulating the abundance and binding affinity of PRPS1 (Tadini et al., 2016). In particular, PRPS1 is the only ribosomal protein that shuttles between ribosome-bound and ribosome-free forms (Merendino et al., 2003; Delvillani et al., 2011), the latter being more abundant in plants that lack GUN1. Based on observations in E. coli, where the unbound form is thought to inhibit translation by competing with ribosomes for mRNAs (Delvillani et al., 2011), it can be argued that the GUN1-dependent equilibrium between the two PRPS1 states has an important role in controlling polysome assembly and protein synthesis in chloroplasts (**Figure 4A**). However, further investigations are needed to clarify this issue. For instance, lines characterized by the ectopic expression of PRPS1 or carrying PRPS1 constructs under the control of inducible promoters, coupled with assays aimed to measure the translation rate in plastids, should allow us to verify the role of PRPS1 in modulating protein synthesis. Furthermore, GUN1 controls the abundance of PRPS1 at the post-transcriptional level. This suggests the involvement of an as yet unidentified plastid protease in this aspect of GUN1 function. In addition, the significance of the interaction of GUN1 with other ribosomal proteins, factors involved in ribosome biogenesis and regulators of plastid protein synthesis remains to be elucidated.

### Chloroplast Protein Import and Degradation

Based on the observations reported above, it appears that GUN1 may well control the interactions of a sub-set of chaperones, promoting plastid protein import when their association with the TIC complex is favored, and stimulating protein degradation, folding/unfolding when they interact with proteases or other protein complexes in the stroma or in the thylakoid membranes (**Figure 4B**). Such a regulatory mechanism would enable GUN1 to coordinate protein translocation across the chloroplast envelope with protein degradation in the stroma,

as well as with plastid division, thus modulating the protein content of the chloroplast in accordance with physiological requirements.

Relatively simple biochemical analyses can be used to verify the importance of GUN1 in influencing the interactions of the stromal chaperones, such as protein complex fractionation via sucrose-gradient ultracentrifugation and/or Blue-Native PAGE coupled with two-dimensional (2D) SDS-PAGE, and immunoblot analyses. Furthermore, the interactions of GUN1 with chaperones should be shown to occur at the plastid envelope and protein import efficiency should be tested in chloroplasts isolated from gun1 and WT seedlings in order to implicate GUN1 in regulating plastid protein import.

### Retrograde Signal Induction

GUN1 may well be a master regulator of plastid-to-nucleus communication in A. thaliana, as it appears to integrate signals derived from perturbations in PGE, TPB, and redox state, in order to modulate nuclear gene expression. Indeed, components of all three pathways have been shown to interact with GUN1, suggesting that signal integration might take place through physical interaction.

Due to the limited abundance of GUN1, as indicated by the fact that the protein has yet to be detected in plastid proteome studies, it is tempting to disregard the idea that its physical interaction with PGE-, TPB-, and redox-related proteins could lead to protein sequestration and directly to differences in protein translation, TPB, and redox balance (Koussevitzky et al., 2007; Pogson et al., 2008; Woodson and Chory, 2008; Kleine and Leister, 2016). Nevertheless, a direct association with GUN1 could control protein abundance through post-transcriptional mechanisms, as in the case of PRPS1 and CHLD (Tadini et al., 2016). Thus, control of CHLD and possibly of FC1 levels could alter the tetrapyrrole flux and influence the abundance of the tetrapyrrole intermediate Mgprotoporphyrin IX (Mg-ProtoIX), or the tetrapyrrole product Fe-protoporphyrin IX (heme), which have been reported to act as negative and positive retrograde signals, respectively (for a review, see Chan et al., 2016). Alternatively, the interaction of GUN1 with the near-identical paralogs ClpC1 and ClpC2 could contribute to the coordination of plastid protein content with tetrapyrrole biosynthesis. Indeed, the activity of the stromal Clp protease has been shown to modulate tetrapyrrole flux by controlling (i) the accumulation of chlorophyll a oxygenase, which converts chlorophyll a into chlorophyll b (Nakagawara et al., 2007), and (ii) the level of glutamyl-tRNA reductase (GluTR), thus regulating the ratelimiting reaction in tetrapyrrole synthesis—the conversion of glutamate-1-semialdehyde into 5-aminolevulinic acid (Apitz et al., 2016).

Therefore, accurate determination of tetrapyrrole intermediates should be performed in gun1 mutant and WT backgrounds. The analyses should be restricted to young seedlings or even to different developmental stages of the chloroplast, in line with the roles of tetrapyrrole and GUN1-mediated signaling in chloroplast development.

### CONCLUDING REMARKS

In the past decade, substantial progress has been made in elucidating retrograde signaling, with the identification of multiple retrograde pathways and more than 40 components involved at different levels in chloroplast-to-nucleus communication. Nevertheless, the molecular function of GUN1 has remained unclear until the recent identification of the GUN1 protein's partners. Based on the functional roles of GUN1 interactors and the embryo lethal or albino phenotypes of most of the corresponding knock-out mutants, we have learned that GUN1 plays a role in chloroplast biogenesis, possibly by controlling protein turnover and protein import, and through the coordination of plastid and nuclear gene expression. Furthermore, GUN1 could have a role in the cpUPR process. Nonetheless, the involvement of GUN1 in plastid biogenesis and protein homeostasis is only just beginning to be understood. For instance, other approaches will be needed to validate the GUN1's protein partners identified by CoIP-MS. The use of a GUN1-GFP protein chimera, expressed under the control of a strong constitutive promoter such as the Cauliflower Mosaic Virus 35S (35S-CaMV), is indeed prone to the identification of false interactors. CoIP-MS studies using a GUN1 specific antibody appears to be the ideal strategy to identify protein partners. Alternatively, the use of GUN1 chimeras under the control of GUN1 native promoter is also practicable. Moreover, we do not know whether all these activities take place within one GUN1-containing nucleoid or if there are different nucleoids/locations for each GUN1-dependent function. The developmental stages of the chloroplast itself may even show distinct patterns of compartmentalization of the different functions. In addition, GUN1's interactions with its diverse partners might have quite different functional consequences: (i) promote specific functions, by bringing enzymes into close proximity with their own substrates and, ultimately, controlling the enzyme abundance, (ii) inhibit processes by sequestering sub-pools of specific proteins and, also in this case, controlling their abundance.

We are confident that future work, based on the exciting breakthroughs discussed in this Review, will shed new light on the molecular functions of GUN1 and its involvement in chloroplast biogenesis and protein homeostasis.

### AUTHOR CONTRIBUTIONS

MC, LT, CP, RF, and PP participated to the organization of the manuscript. MC and PP designed and conceived the pictures. PP wrote the manuscript.

### ACKNOWLEDGMENTS

This work was supported by ERA-NET Cofund FACCE SURPLUS (BarPLUS grant id. 93). We thank Paul Hardy for critical reading of the manuscript. The authors wish to acknowledge the reviewers for the detailed and helpful comments to the manuscript.

### REFERENCES


regulatory feedback loops in chlamydomonas. Plant Cell 25, 167–186. doi: 10.1105/tpc.112.103051


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2016 Colombo, Tadini, Peracchio, Ferrari and Pesaresi. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

## Ubiquitin–Proteasome-Dependent Regulation of Bidirectional Communication between Plastids and the Nucleus

#### Yoshihiro Hirosawa<sup>1</sup> , Yasuko Ito-Inaba1,2 and Takehito Inaba<sup>1</sup> \*

<sup>1</sup> Department of Agricultural and Environmental Sciences, Faculty of Agriculture, University of Miyazaki, Miyazaki, Japan, <sup>2</sup> Organization for Promotion of Tenure Track, University of Miyazaki, Miyazaki, Japan

Plastids are DNA-containing organelles and can have unique differentiation states depending on age, tissue, and environment. Plastid biogenesis is optimized by bidirectional communication between plastids and the nucleus. Import of nuclear-encoded proteins into plastids serves as anterograde signals and vice versa, plastids themselves send retrograde signals to the nucleus, thereby controlling de novo synthesis of nuclear-encoded plastid proteins. Recently, it has become increasingly evident that the ubiquitin–proteasome system regulates both the import of anterograde plastid proteins and retrograde signaling from plastids to the nucleus. Targets of ubiquitin–proteasome regulation include unimported chloroplast precursor proteins in the cytosol, protein translocation machinery at the chloroplast surface, and transcription factors in the nucleus. This review will focus on the mechanism through which the ubiquitin–proteasome system optimizes plastid biogenesis and plant development through the regulation of nuclear–plastid interactions.

#### Edited by:

Dario Leister, Ludwig Maximilian University of Munich, Germany

#### Reviewed by:

Karin Krupinska, University of Kiel, Germany Sacha Baginsky, Martin Luther University of Halle-Wittenberg, Germany

\*Correspondence: Takehito Inaba tinaba@cc.miyazaki-u.ac.jp

#### Specialty section:

This article was submitted to Plant Cell Biology, a section of the journal Frontiers in Plant Science

Received: 13 December 2016 Accepted: 20 February 2017 Published: 15 March 2017

#### Citation:

Hirosawa Y, Ito-Inaba Y and Inaba T (2017) Ubiquitin–Proteasome-Dependent Regulation of Bidirectional Communication between Plastids and the Nucleus. Front. Plant Sci. 8:310. doi: 10.3389/fpls.2017.00310 Keywords: proteasome, ubiquitin, plastid protein import, plastid biogenesis, retrograde signaling

Plastids are DNA-containing organelles that have evolved from a cyanobacterial endosymbiont. Because most of the genes encoded by the bacterial ancestor have been transferred to the host nuclear genome, the expression of genes encoding plastid precursor proteins in the nucleus and the import of those proteins are indispensable for plastid biogenesis. The import of plastid precursor proteins is mediated by the translocon at the outer (or inner) envelope membrane of chloroplasts (TOC-TIC). Hence, plastid fate is largely controlled by the quality and quantity of plastid precursor proteins expressed in each cell and by their import into plastids (Inaba and Schnell, 2008; Jarvis and Lopez-Juez, 2013; Paila et al., 2015).

On the other hand, plastids also send feedback signals to regulate the expression of genes encoding plastid proteins in the nucleus. These signals are known as retrograde signals from plastids to the nucleus and are referred to as plastid signals. Plastid signals can be divided into two types: biogenic and operational (Pogson et al., 2008). Among them, biogenic signals are necessary to coordinate gene expression in two genomes, allowing cells to assemble the photosynthetic apparatus and to promote chloroplast development (Pogson et al., 2008; Inaba et al., 2011; Jarvis and Lopez-Juez, 2013). To date, several transcription factors have been shown to mediate biogenic signals from plastids to the nucleus (Koussevitzky et al., 2007; Ruckle et al., 2007; Kakizaki et al., 2009; Waters et al., 2009; Sun et al., 2011; Martin et al., 2016).

A number of studies have demonstrated the roles of de novo synthesis and the targeting of plastid precursor proteins in the regulation of nuclear–plastid interactions. However, it has become increasingly evident that the nuclear–plastid interaction is also regulated by the degradation of multiple components through the ubiquitin–proteasome system (Lee et al., 2013; Ling and Jarvis, 2015). Here, we focus on recent advances in our understanding

of how the ubiquitin–proteasome system regulates the nuclear– plastid interaction and plastid biogenesis. Other comprehensive reviews cover broad aspects of plastid protein import and plastid signaling (Li and Chiu, 2010; Inaba et al., 2011; Jarvis and Lopez-Juez, 2013; Pfannschmidt and Munné-Bosch, 2013; Paila et al., 2015; Chan et al., 2016), and space limitations prevent us from providing adequate coverage of all aspects of nuclear–plastid interaction.

### DEGRADATION OF UNIMPORTED CHLOROPLAST PRECURSOR PROTEINS BY THE UBIQUITIN–PROTEASOME PATHWAY

It is well known that the expression of nuclear-encoded photosynthesis-associated genes are induced upon illumination and that mass transport of proteins encoded by these genes into plastids are indispensable for chloroplast development. Those plastid-targeted proteins are encoded as precursors in the nucleus, but only mature proteins are detectable under normal conditions in vivo. To avoid the accumulation of unimported proteins in the cytosol, plants have evolved at least two distinct mechanisms. One is feedback regulation of nuclear gene expression by plastid-derived signals, and the other is degradation of unimported precursor proteins by the ubiquitin– proteasome system (Lee et al., 2013; **Figure 1**).

Cytosolic heat shock cognate 70-4 (Hsc70-4) and carboxy terminus of Hsc70-interacting protein (CHIP) appear to be involved in the degradation of unimported precursor proteins in Arabidopsis thaliana (Lee et al., 2009). Hsc70-4 recognizes specific sequence motifs within the transit peptide of ribulose-1,5-bisphosphate carboxylase/oxygenase small subunit protein and light-harvesting chlorophyll a/b-binding protein. Subsequently, CHIP interacts with Hsc70-4 and serves as an E3-ubiquitin ligase, thereby allowing unimported precursors to be degraded through the ubiquitin–proteasome system. This suggests that a transit peptide may function as both a chloroplast targeting signal and a degradation signal when unimported precursors accumulate in the cytosol. This idea is further substantiated by the findings of a recent proteomic study (Sako et al., 2014), in which certain plastid precursors were shown to interact with the proteasome both in vivo and in vitro.

The mechanism that discriminates between plastid-targeted precursors and proteasome-targeted precursors remains to be characterized in detail. Intriguingly, the plastid protein import 2 (ppi2) mutant of A. thaliana, which lacks the atToc159 protein import receptor of plastids, accumulated N-acetylated plastid precursor proteins outside of plastids (Bischof et al., 2011). Although atToc159 plays key roles in the import of photosynthesis-associated proteins into plastids, it also participates in the import of constitutively expressed plastid proteins. As will be discussed later, the ppi2 mutant has been known to exhibit down-regulation of genes encoding photosynthesis-associated proteins in the nucleus in response to plastid-derived signals, but not the expression of constitutive plastid proteins. In contrast, some constitutively expressed proteins were shown to be N-acetylated in the ppi2 mutant. It has been shown that N-acetylation serves as a degradation signal for the ubiquitin–proteasome system in yeast (Hwang et al., 2010). Hence, one can speculate that excess precursors that cannot be controlled at the transcriptional level are subjected to N-acetylation and ubiquitin–proteasome-dependent degradation. As such, degradation of excess plastid precursors via the ubiquitin–proteasome system plays a key role in determining the amount of protein import and plastid biogenesis.

### PLASTID PROTEIN IMPORT MACHINERY IS A DIRECT TARGET OF UBIQUITIN–PROTEASOME PATHWAY

The ubiquitin–proteasome system directly regulates the protein translocation machinery at the plastid surface (Ling and Jarvis, 2015; **Figure 1**). This unexpected link was demonstrated in an attempt to isolate a suppressor mutant of plastid protein import 1 (ppi1). The ppi1 mutant of A. thaliana exhibits a pale green phenotype due to the lack of atToc33 in the TOC complex, but can survive on soil. One suppressor mutant of ppi1, designated as suppressor of ppi1 locus1 (sp1), possesses a lesion within the RING-type ubiquitin E3 ligase gene (Ling et al., 2012). TOC components are more abundant (1.5- to 2-fold) in the sp1 mutant than in the wild-type. The wild-type SP1 protein interacts with components of TOC machinery. Furthermore, atToc159, atToc75, and atToc33 have been shown to be polyubiquitinated by SP1 activity. These findings indicate that the ubiquitin–proteasome system directly regulates the level of TOC components, thereby affecting the amount of protein import into plastids.

This mechanism also seems to play a key role in determining the fate of plastids within the cell (Ling et al., 2012). During the photomorphogenic response, the sp1 single mutant displayed inefficient de-etiolation with reduced levels of photosynthesisassociated proteins and imbalanced TOC receptor levels. The sp1 mutant also exhibited delayed senescence, and this was accompanied by the delayed transformation from chloroplasts to gerontoplasts within the cell. In contrast, overexpression of SP1 accelerated both de-etiolation and senescence. Hence, regulation of TOC components by the ubiquitin–proteasome system appears to be indispensable for determining both the quality and the quantity of plastidtargeted proteins, thereby affecting the fate of plastid and plant development.

### REGULATION OF PLASTID-TO-NUCLEUS RETROGRADE SIGNALING VIA THE UBIQUITIN–PROTEASOME PATHWAY

In addition to the anterograde signaling pathway, a recent study demonstrated that the retrograde signaling pathway from

plastids to the nucleus is also subjected to ubiquitin–proteasomedependent regulation in A. thaliana (Tokumaru et al., 2017). The key mechanism involves the regulation of the GOLDEN2-LIKE 1 (GLK1) transcription factor by the ubiquitin–proteasome system (**Figure 1**).

The GLK family of transcription factors was originally isolated in maize (Hall et al., 1998; Rossini et al., 2001). The GLK genes positively regulate the expression of photosynthesisassociated genes in numerous plants, thereby strongly promoting chloroplast development (Fitter et al., 2002; Yasumura et al., 2005;

Waters et al., 2009). Overexpression of GLK has been shown to be sufficient to induce chloroplast development in rice calli (Nakamura et al., 2009) and Arabidopsis root cells (Kobayashi et al., 2012; Tokumaru et al., 2017). Two separate studies reported that the expression of GLK genes responds to inhibitor treatment thus compromising chloroplast development (Kakizaki et al., 2009; Waters et al., 2009). The findings of those studies concluded that GLK gene expression responds to plastid signals, resulting in the regulation of photosynthesis-associated genes in response to plastid signals. Intriguingly, impaired chloroplast development caused by the ppi2 mutation also suppress GLK1 expression in the nucleus (Kakizaki et al., 2009). This regulation is mediated by the retrograde signaling pathway, because the GENOMES UNCOUPLED 1 (GUN1) protein, which is localized in plastids, appears to act upstream of GLK1. From those studies, it becomes clear that plastids transmit signals to determine the amount of anterograde protein import, thereby avoiding the accumulation of excess levels of precursors within the cytosol.

Besides transcriptional regulation, a recent study showed that plastid signals also directly regulate the level of GLK1 protein (Tokumaru et al., 2017). The GLK1 gene is fully expressed in gun1 mutants treated with norflurazon. In contrast, the level of GLK1 protein is much lower than that expected from the GLK1 mRNA levels in the norflurazon-treated gun1 mutant. The discrepancy between GLK1 protein and mRNA levels is in part attributable to the degradation of the GLK1 protein by the ubiquitin–proteasome system (Tokumaru et al., 2017). When norflurazon-treated plants were further treated with MG-132, a proteasome inhibitor, the accumulation of GLK1 was partially restored. Because the gun1 mutant also exhibited the same response, it appears that GUN1 is not required for the proteasome-mediated regulation of GLK1. Likewise, MG-132 treatment partially restored the level of GLK1 protein in the ppi2 mutant. Hence, this mechanism is also used to optimize the expression of nuclear genes encoding photosynthesis-associated proteins when plastid protein import is compromised.

The level of GLK is also regulated by the ubiquitin– proteasome system in fruit tissues of tomato. The Solanum lycopersicum GLK2 protein, SlGLK2, regulates chloroplast development in tomato fruit tissues, and fruits of the slglk2 mutant exhibit uniformly light green coloration (Powell et al., 2012; Nguyen et al., 2014). SlGLK2 was found to be degraded by the ubiquitin E3 ligase complex containing CULLIN4 (CUL4) and UV-DAMAGED DNA BINDING PROTEIN 1 (DDB1; Tang et al., 2016). Consistent with this observation, a mutation

### REFERENCES


in DDB1 significantly increased the pigment contents and chloroplast/chromoplast size in tomato fruits (Cookson et al., 2003), presumably due to the excess accumulation of SlGLK2. Although the roles of SlGLK2 in retrograde signaling remains to be established, these studies further support the idea that the ubiquitin–proteasome pathway is indispensable for the regulation of GLKs.

### CONCLUSION AND PERSPECTIVE

Although the de novo synthesis and targeting of plastid precursor proteins are indispensable for plastid biogenesis, it becomes clear that ubiquitin–proteasome-dependent protein degradation also plays a key role in the regulation of plastid biogenesis. Meanwhile, a number of questions remains to be solved: Are there any other ubiquitin–proteasome regulated transcription factors involved in the retrograde signaling from plastids to the nucleus? Is ubiquitin–proteasome system indispensable for the retrograde signaling from plastids other than chloroplasts? Does operational control of retrograde signaling also requires ubiquitin–proteasome system? In fact, other studies start addressing these questions. Proteasome-regulated transcription factors, such as ELONGATED HYPOCOTYL 5 (HY5) and PHYTOCHROME INTERACTING FACTORS (PIFs), have been shown to participate in retrograde signaling, as well as in light signaling (Ruckle et al., 2007; Martin et al., 2016). Reactive oxygen species-producing chloroplasts appear to be ubiquitinated and subsequently degraded (Woodson et al., 2015). Further investigation will provide novel insight into the roles of the ubiquitin–proteasome system in regulating plastid biogenesis and plant development.

### AUTHOR CONTRIBUTIONS

TI wrote the manuscript with assistance of YH and YI-I.

### FUNDING

This work was supported in part by Grants-in-Aid for Scientific Research (15K07843), Grants-in-Aid for Young Scientists (B, No. 26850065), the Program to Disseminate Tenure Tracking System from the Japanese Ministry of Education, Culture, Sports, Science, and Technology.

phytoene synthase activity underlie the phenotype of the high pigment-1 mutant of tomato. Planta 217, 896–903. doi: 10.1007/s00425-003-1065-9



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2017 Hirosawa, Ito-Inaba and Inaba. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.